Electronic watermark embedding method, device, and program, and electronic watermark detecting method, device, and program
Abstract
An electronic watermark embedding device for embedding embedment information as an electronic watermark in an input signal having N or more dimensions (N is 2 or more an integer) and an electronic watermark detecting device are disclosed. The electronic watermark embedding device creates an embedment sequence according to embedment information, creates an (N-1)-dimensional pattern according to the embedment sequence, modulates a periodical signal according to the value on the (N-1)-dimensional pattern, thus creates an N-dimensional embedment pattern, superimposes the embedment pattern on the input signal, and outputs the resultant signal. The electronic watermark detecting device measures the component of a predetermined periodic signal in one dimensional direction of the input signal, obtains an (N-1)-dimensional pattern, obtains a detection sequence from the values of the (N-1)-dimensional pattern, and detects the embedded electronic watermark according to the magnitude of the value of the correlation between the detection sequence and the embedded sequence.

Term
No projected expiry on record.
- Priority
- Filed
- Published
- Today
38 claims: 35 independent, 3 dependent
- 1請求の範囲 [1] 埋め込み系列生成手段、配列生成手段、変調手段、記憶手段、埋め込みパターン 重畳手段と、を有する電子透力 埋め込み装置において、 N (Nは 2以上の整数)以 上の次元を持つ入力信号に対して埋め込み情報を電子透力しとして人間の知覚に 感知されな!、ように埋め込む電子透かし埋め込み方法であって、 前記埋め込み系列生成手段が、前記埋め込み情報に基づき埋め込み系列を生成 し、第 1の記憶手段に格納し、 前記配列生成手段が、前記第 1の記憶手段の前記埋め込み系列に基づき N— 1次 元パターンを生成し、 前記変調手段が、前記 N— 1次元パターン上の値に応じて周期信号を変調するこ とにより N次元の埋め込みパターンを生成し、第 2の記憶手段に格納し、 前記埋め込みパターン重畳手段が、前記第 2の記憶手段に格納されて 、る前記 N 次元の埋め込みパターンを取得し、該埋め込みパターンを前記入力信号に重畳す る埋め込む、 ことを特徴とする電子透かし埋め込み方法。
- 2[2] 前記変調手段は、前記 N— 1次元パターン上の位置によって N次元目の方向の位 相がそれぞれ異なるように N次元の埋め込みパターンを生成する請求項 1記載の電 子透かし埋め込み方法。
- 3[3] 前記 N— 1次元パターンが複素数パターンであって、 前記配列生成手段は、前記埋め込み系列の一部を実部、一部を虚部となるように 前記 N— 1次元パターンを生成する請求項 1または 2記載の電子透かし埋め込み方 法。
- 4[4] 前記 N— 1次元パターンが複素数パターンであって、 前記変調手段は、前記 N— 1次元パターン上の複素数の偏角が変調信号の位相と なり、絶対値が変調信号の大きさとなるように前記周期信号を変調する請求項 1乃至 3のうちいずれか 1項記載の電子透かし埋め込み方法。
- 5[5] 前記埋め込み系列生成手段が、生成された前記埋め込み系列を分割して複数の 埋め込み系列を生成し、前記第 1の記憶手段に格納し、 前記配列生成手段が、前記第 1の記憶手段に格納された前記複数の埋め込み系 列毎に各々対応する N— 1次元パターンを生成し、 前記変調手段が、前記 N— 1次元パターン毎に各々対応する N次元の埋め込みパ ターンを生成し、前記第 2の記憶手段に格納し、 前記埋め込みパターン重畳手段が、前記第 2の記憶手段の前記埋め込みパターン を全て加算した後に、前記入力信号に重畳する、 請求項 1乃至 4のうちいずれか 1項記載の電子透力し埋め込み方法。
- 6[6] 埋め込み系列生成手段、配列生成手段、変換手段、記憶手段、埋め込みパターン 重畳手段、逆変換手段と、を有する電子透かし埋め込み装置において、 N (Nは 2以 上の整数)以上の次元を持つ入力信号に対して埋め込み情報を電子透力しとして人 間の知覚に感知されないように埋め込む電子透かし埋め込み方法であって、 前記埋め込み系列生成手段が、前記埋め込み情報に基づき埋め込み系列を生成 し、第 1の記憶手段に格納し、 前記配列生成手段が、前記第 1の記憶手段に格納されている前記埋め込み系列 に基づき N— 1次元パターンを生成し、第 2の記憶手段に格納し、 前記変換手段が、前記入力信号を直交変換し、変換済信号を取得し、 前記埋め込みパターン重畳手段が、前記第 2の記憶手段に格納されて 、る前記 N 1次元パターンを前記変換済信号の一部の N— 1次平面に重畳し、逆変換前信号 を取得し、 前記逆変換手段が、前記逆変換前信号を直交逆変換し、埋め込み済み信号を得 る、 ことを特徴とする電子透かし埋め込み方法。
- 7[7] 前記埋め込み系列生成手段が、複数の埋め込み系列を生成して前記第 1の記憶 手段に格納し、 前記配列生成手段が、前記第 1の記憶手段に格納されている前記複数の埋め込 み系列毎に各々対応する N— 1次元パターンを生成し、前記第 2の記憶手段に格納 し、 前記埋め込みパターン重畳手段が、前記第 2の記憶手段に格納されて 、る前記 N 1次元パターンを前記変換済み信号の複数の N— 1次元平面にそれぞれ重畳す る、請求項 6記載の電子透かし埋め込み方法。
- 8[8] 復調手段、検出系列抽出手段、相関値計算手段、記憶手段と、を有する電子透か し検出装置にぉ 、て、 N (Nは 2以上の整数)以上の次元を持つ入力信号に対して予 め人間の知覚に感知されないように埋め込まれた電子透力 を検出する電子透かし 検出方法であって、 前記復調手段が、前記入力信号の一つの次元方向における所定の周期信号の成 分を測定し、 N— 1次元パターンを求め、 前記検出系列抽出手段が、前記 N— 1次元パターンの値力 検出系列を求め、記 憶手段に格納し、 前記相関値計算手段が、前記記憶手段に格納された前記検出系列と埋め込み系 列の相関値の大きさに基づ 、て、埋め込まれて!/ヽる電子透かしを検出する、 ことを特徴とする電子透かし検出方法。
- 9[9] 前記復調手段が、同一の周波数を持つ直交する 2つの周期信号を生成し、前記入 力信号と前記周期信号との相関に基づ!/、て N - 1次元パターンを求める、請求項 8 記載の電子透かし検出方法。
- 10[10] 前記復調手段が、前記入力信号を N次元目の方向の差分もしくは微分値に基づい て復調する、請求項 8または 9記載の電子透かし検出方法。
- 11[11] 前記 N— 1次元パターンが複素数パターンであって、 前記検出系列抽出手段が、前記 N— 1次元パターンの実部及び虚部の値に基づ いて前記検出系列を求め、前記記憶手段に格納する、請求項 8乃至 10のうちいず れか 1項記載の電子透かし検出方法。
- 12[12] 前記 N— 1次元パターンが複素数パターンであって、 前記相関値計算手段が、各ビット毎の複素相関値を求め、各ビット毎の複素相関 値の向きを揃えた上でそれらの総和をとり、この総和に基づいて埋め込まれている電 子透力 を検出する請求項 8乃至 11のうちいずれか 1項記載の電子透かし検出方法
- 13[13] 前記 N— 1次元パターンが複素数パターンであって、 前記相関値計算手段が、複素相関値の絶対値に基づいて埋め込まれている電子 透力 を検出する、請求項 8乃至 11のうちいずれ力 1項記載の電子透かし検出方法
- 14[14] 前記 N— 1次元パターンが複素数パターンであって、 前記相関値計算手段が、各ビット毎の複素相関値を求め、各ビット毎の複素相関 値の向きを揃えた上でそれらの総和をとり、同期手段が、この総和の偏角に基づき、 前記入力信号の同期変位量を求める請求項 8乃至 13のうちいずれか 1項記載の電 子透かし検出方法。
- 15[15] 前記 N— 1次元パターンが複素数パターンであって、 同期手段が、前記検出系列と前記埋め込み系列の複素相関値の偏角に基づき、 前記入力信号の同期変位量を求める請求項 8乃至 13のうちいずれか 1項記載の電 子透かし検出方法。
- 16[16] 前記復調手段が、複数の周期信号の位相を測定し、複数の N— 1次元パターンを 求め、 前記同期手段が、複数の N— 1次元パターン毎に各々同期変位量を求め、 前記検出系列抽出手段が、前記複数の N— 1次元パターンを各々対応する前記同 期変位量に基づいて同期を補正して検出系列を求め、前記記憶手段に格納し、 前記相関値計算手段が、前記複数の N— 1次元パターン毎に得られた前記記憶手 段に格納されている前記検出系列を結合した系列と前記埋め込み系列の相関値を 計算する、 請求項 15記載の電子透かし検出方法。
- 17[17] 前記同期手段は、前記複数の N— 1次元パターン毎に得られた同期変位量を元に 、 N次元目の軸方向の全体の変位量を求める、請求項 16記載の電子透かし検出方 法。
- 18[18] 前記同期手段は、予め埋め込まれた同期系列を検出し、同期合わせを行い、前記 同期変位量に応じて入力信号を再分割し、残る複数の埋め込み情報を検出する請 求項 15記載の電子透かし検出方法。
- 19[19] N (Nは 2以上の整数)以上の次元を持つ入力信号に対して埋め込み情報を電子透 力しとして人間の知覚に感知されないように埋め込む電子透力し埋め込み装置であ つて、 前記埋め込み情報に基づき埋め込み系列を生成し、第 1の記憶手段に格納する埋 め込み系列生成手段と、 前記第 1の記憶手段の前記埋め込み系列に基づき N— 1次元パターンを生成する 配列生成手段と、 前記 N— 1次元パターン上の値に応じて周期信号を変調することにより N次元の埋 め込みパターンを生成し、第 2の記憶手段に格納する変調手段と、 前記第 2の記憶手段に格納されている前記 N次元の埋め込みパターンを取得し、 該埋め込みパターンを前記入力信号に重畳する埋め込みパターン重畳手段と、 を有することを特徴とする電子透かし埋め込み装置。
- 20[20] 前記変調手段は、 前記 N— 1次元パターン上の位置によって N次元目の方向の位相がそれぞれ異な るように N次元の埋め込みパターンを生成する請求項 19記載の電子透力し埋め込 み装置。
- 21[21] 前記 N— 1次元パターンが複素数パターンであって、 前記配列生成手段は、前記埋め込み系列の一部を実部、一部を虚部となるように 前記 N— 1次元パターンを生成する請求項 19または 20記載の電子透かし埋め込み 装置。
- 22[22] 前記 N— 1次元パターンが複素数パターンであって、 前記変調手段は、前記 N— 1次元パターン上の複素数の偏角が変調信号の位相と なり、絶対値が変調信号の大きさとなるように前記周期信号を変調する請求項 19乃 至 21のうちいずれか 1項記載の電子透かし埋め込み装置。
- 23[23] 前記埋め込み系列生成手段は、生成された前記埋め込み系列を分割して複数の 埋め込み系列を生成し、前記第 1の記憶手段に格納し、 前記配列生成手段は、前記第 1の記憶手段に格納された前記複数の埋め込み系 列毎に各々対応する N— 1次元パターンを生成し、 前記変調手段は、前記 N— 1次元パターン毎に各々対応する N次元の埋め込みパ ターンを生成し、前記第 2の記憶手段に格納し、 前記埋め込みパターン重畳手段は、前記第 2の記憶手段の前記埋め込みパター ンを全て加算した後に、前記入力信号に重畳する、 請求項 19乃至 22のうちいずれか 1項記載の電子透力し埋め込み装置。
- 24[24] N (Nは 2以上の整数)以上の次元を持つ入力信号に対して埋め込み情報を電子透 力しとして人間の知覚に感知されないように埋め込む電子透力し埋め込み装置であ つて、 前記埋め込み情報に基づき埋め込み系列を生成し、第 1の記憶手段に格納する埋 め込み系列生成手段と、 前記第 1の記憶手段に格納されている前記埋め込み系列に基づき N— 1次元バタ ーンを生成し、第 2の記憶手段に格納する配列生成手段と、 前記入力信号を直交変換し、変換済信号を得る変換手段と、 前記第 2の記憶手段に格納されている前記 N— 1次元パターンを前記変換済信号 の一部の N— 1次平面に重畳し、逆変換前信号を得る埋め込みパターン重畳手段と 前記逆変換前信号を直交逆変換し、埋め込み済み信号を得る逆変換手段と、 を有することを特徴とする電子透かし埋め込み装置。
- 25[25] 前記埋め込み系列生成手段は、複数の埋め込み系列を生成して前記第 1の記憶 手段に格納し、 前記配列生成手段は、前記第 1の記憶手段に格納されている前記複数の埋め込 み系列毎に各々対応する N— 1次元パターンを生成し、前記第 2の記憶手段に格納 し、 前記埋め込みパターン重畳手段は、前記第 2の記憶手段に格納されて 、る前記 N 1次元パターンを前記変換済み信号の複数の N— 1次元平面にそれぞれ重畳す る、請求項 24記載の電子透かし埋め込み装置。
- 26[26] N (Nは 2以上の整数)以上の次元を持つ入力信号に対して予め人間の知覚に感 知されな 、ように埋め込まれた電子透力 を検出する電子透かし検出装置であって、 前記入力信号の一つの次元方向における所定の周期信号の成分を測定し、 N— 1 次元パターンを求める復調手段と、 前記 N— 1次元パターンの値力 検出系列を求め、記憶手段に格納する検出系列 抽出手段と、 前記記憶手段に格納された前記検出系列と埋め込み系列の相関値の大きさに基 づいて、埋め込まれている電子透力しを検出する相関値計算手段と、 を有することを特徴とする電子透かし検出装置。
- 27[27] 前記復調手段は、同一の周波数を持つ直交する 2つの周期信号を生成し、前記入 力信号と前記周期信号との相関に基づ!/、て N - 1次元パターンを求める請求項 26 記載の電子透かし検出装置。
- 28[28] 前記復調手段は、前記入力信号を N次元目の方向の差分もしくは微分値に基づい て復調を行う、請求項 26または 27記載の電子透かし検出装置。
- 29[29] 前記 N— 1次元パターンが複素数パターンであって、 前記検出系列抽出手段は、前記 N— 1次元パターンの実部及び虚部の値に基づ いて前記検出系列を求め、前記記憶手段に格納する、請求項 26乃至 28のうちいず れカ 1項記載の電子透かし検出装置。
- 30[30] 前記 N— 1次元パターンが複素数パターンであって、 前記相関値計算手段が、各ビット毎の複素相関値を求め、各ビット毎の複素相関 値の向きを揃えた上でそれらの総和をとり、この総和に基づいて埋め込まれている電 子透力 を検出する請求項 26乃至 29のうちいずれ力 1項記載の電子透かし検出装 置。
- 31[31] 前記 N— 1次元パターンが複素数パターンであって、 前記相関値計算手段は、複素相関値の絶対値に基づいて埋め込まれている電子 透力 を検出する、請求項 26乃至 29のうちいずれか 1項記載の電子透かし検出装 置。
- 32[32] 前記 N— 1次元パターンが複素数パターンであって、 前記相関値計算手段が、各ビット毎の複素相関値を求め、各ビット毎の複素相関 値の向きを揃えた上でそれらの総和をとり、 この総和の偏角に基づき、前記入力信号の同期変位量を求める同期手段を有する 請求項 26乃至 31のうちいずれか 1項記載の電子透かし検出装置。
- 33[33] 前記 N— 1次元パターンが複素数パターンであって、 前記検出系列と前記埋め込み系列の複素相関値の偏角に基づき、前記入力信号 の同期変位量を求める同期手段を有する、請求項 26乃至 31のうちいずれか 1項記 載の電子透かし検出装置。
- 34[34] 前記復調手段は、複数の周期信号の位相を測定し、複数の N— 1次元パターンを 求め、 前記同期手段は、複数の N— 1次元パターン毎に各々同期変位量を求め、 前記検出系列抽出手段は、前記複数の N— 1次元パターンから各々対応する前記 同期変位量に基づいて同期を補正して検出系列を求め、前記記憶手段に格納し、 前記相関値計算手段は、前記複数の N— 1次元パターン毎に得られた前記記憶手 段に格納されている前記検出系列を結合した系列と前記埋め込み系列の相関値を 計算する、 請求項 33記載の電子透かし検出装置。
- 35[35] 前記同期手段は、 前記複数の N— 1次元パターン毎に得られた同期変位量を元に、 N次元目の軸方 向の全体の変位量を求める請求項 34記載の電子透かし検出装置。
- 36[36] 前記同期手段は、予め埋め込まれた同期系列を検出し、同期合わせを行い、前記 同期変位量に応じて入力信号を再分割し、残る複数の埋め込み情報を検出する検 出手段を有する請求項 34記載の電子透かし検出装置。
- 37[37] N (Nは 2以上の整数)以上の次元を持つ入力信号に対して埋め込み情報を電子透 力しとして人間の知覚に感知されないように埋め込む電子透力し埋め込みプログラム であって、 コンピュータを、 請求項 19乃至 25のうちいずれか 1項記載の電子透かし埋め込み装置として機能さ せることを特徴とする電子透かし埋め込みプログラム。
- 38[38] N (Nは 2以上の整数)以上の次元を持つ入力信号に対して予め人間の知覚に感 知されないように埋め込まれた電子透力 を検出する電子透かし検出プログラムであ つて、 コンピュータを、 請求項 26乃至 36のうちいずれか 1項記載の電子透かし検出装置として機能させる ことを特徴とする電子透かし検出プログラム。
Independent claims38
1,747 paragraphs in 54 sections, as filed
Specification
An electronic watermark embedding method, a device, a program, and digital-watermarking detection A method, a device, and program
Technical field
[0001] The present invention relates to an electronic watermark embedding method, a device, a program, an electronic watermark detecting method, a device, and a program, The signal power in which embedded to input signals including a picture signal so that [ another subinformation ] it might not be perceived, and subinformation was embedded especially is also This. The electronic watermark embedding method, the device, the program, and Electronic penetration which read subinformation It lends and is related with a detecting method, a device, and a program.
Background art
[0002] There is art of performing the protection of copyrights of digital contents, by embedding electronic Power to digital contents as conventional technology. There is art of referring to metadata, such as copyright information about digital contents. Digital contents are photoed with a digital camera, after passing analog media, such as printed matter as an advertisement, and it is digital watermarking. There is also art which acquires the information relevant to an advertisement by reading.
[0003] It is electronic Penetration to a still picture. As the method of embedding, It is the real part and imaginary part of a rectangular conversion field (for example, Fourier transform field) of a picture about the embedding series generated by pseudorandom numbers. Spectrum Expansion which embeds and detects using correlation of an embedding series and a detection series The Scatter type electronic Penetration method is indicated (for example, refer to patent documents 1).
[0004] Also in a picture signal, generally, since a picture signal is recorded as a thing which the frame picture which is a still picture was made to follow, it is applying the Electronic transparent ゝ method for a still picture, and it becomes possible to embed electronic Power. For example, Electricity given in the above-mentioned patent documents 1 Using the child watermark method, digital watermarking common to each frame picture of a picture signal is buried, and it is Included. Electronic Penetration [ as opposed to a picture signal at Mumo ] It is realizable.
[0005] When the still picture in which digital watermarking was embedded is used unjustly, the case where cut off a part of picture and it is used can be considered. Place of a picture where the part was cut off The part cut off in the digital-watermarking detection which does not use Synthesis and an original picture image is the original picture. It is not understood in the portion of a throat whether it is Oh. This means quantity with an arbitrary pattern of embedded digital watermarking, and seeming to carry out parallel translation. Namely, this is in the direction of space. The state where the synchronization of the digital-watermarking pattern has shifted is meant. When calling this "a space synchronization of electronic Power" and detecting electronic Penetration, it is necessary to unite a space synchronization by methods , such as clarifying the amount of parallel translation (although the amendment to geometric modification of Affine strange Conversion etc. may generally also be included in the space synchronization of electronic Power). In the present invention, it is parallel translation. It is aimed at the receiving amendment.
[0006] A picture signal is treated as a set of a plurality of still pictures (frame) which generally stand in a row in the direction of time. In the electronic Penetration method for video, it is one among sets of such a frame. Set power peach electronic Power of the frame which a portion follows is wanted to be detectable. for example, -- carrying out by starting only one scene of the distributed image contents and being used unjustly Watch -- only as for the scene by which the illegal use was carried out also when like, power can also detect electronic Power -- un--- The effect of the deterrence for Masatoshi is expectable. Contest an image currently projected in a movie theater etc., for example When re-photography image Power ゝ electronic PenetrationゝTo photoed with a video camera in Ten is detected The starting point of an image at the time of embedding of electronic Power and the starting point of a photoed image are To inevitably. It has been., A case is also expected for electronic Power to be detectable. For example, cameras, such as a personal digital assistant, are used from the scene of image contents, just displayed now. The photoed image power is also electronic Penetration. Application which detects and acquires pertinent information It thinks. which portion of the image where electronic Power was embedded in these examples Are started! or [ /and To ] -- beforehand -- Component force -- and a sake -- detection Well and The of electronic Power -- candidate for detection the necessity of knowing to which position of the signal with which the part carried out electronic With power and was embedded it corresponding -- is there. This will be called the time synchronization of electronic Power.
[0007] The space synchronization in the conventional digital-watermarking method and a time synchronization method can roughly be classified as follows.
[0008] (One) comprehensive search : it is detection of electronic Power respectively to all the amounts of synchronous displacement considered. It tries comprehensively one by one.
[0009] Embedding of the signal for (2) matching synchronization : embed the signal for matching synchronization aside from electronic Power, and unite a synchronization by detecting it. [0010] For example, electronic Penetration given in the above-mentioned patent documents 1 By a method, Digital watermarking is spatial. The signal for detecting the amount of parallel translation is embedded, and it embeds in piles with information, and is this Trust. Comprehensive search of the amount of synchronous displacement of an item is efficiently carried out using a discrete Fourier transform. Space matching synchronization is performed.
Patent documents 1 : Provisional publication of a patent 2003 -- No. 219148 gazette
The indication of an invention
Object of the Invention
[0011] With how for the former to carry out electronic Penetration for images, and embed [ carry out, and ] while carrying out power, the following subjects are Oho.
[0012] Subject of the tolerance to - quantity compression or re-photography:
For example, MPEG1/2/4, and WMV (Windows (registered trademark) Media Video), Video irreversible numerals Y of a high compression ratio, such as DiVX and H.264/A VC, are applied (high compression), Screen It is [ a video camera and ] carrying again about the image outputted to a display device called a display. Detection is difficulty when a photograph is taken using the camera etc. which were carried in the telephone (re-photography). There was a case.
[0013] It is electronic Penetration in order to make detection possible also by high compression or re-photography. It will be necessary to embed strongly and they are Invitation , The , and It was about deterioration of the image quality of an image as a result.
[0014] In order to give sufficient tolerance for quantity compression, re-photography, etc., maintaining sufficient image quality conversely, the information length embedded on an image needed to be shortened.
[0015] Subject of diffuse series length:
In literature "Kana Yamamoto, Takao Nakamura, Yoichi Takashima, Jun Katayama, Ryo Kitahara, and * Taketaka "1 about detection quality assessment of frame superposition type Motion picture electronic Power consideration" information-science-and-technology forum FIT2005, J-029, and 2005", The reliability of detection of electronic Power is high Up by lengthening and the diffuse series length of spectral diffusion in the electronic Penetration method using spectral diffusion and correlation computation. To is described. It embeds in the electronic Penetration method of the above-mentioned patent documents 1. A real value the number of the frequency coefficient positions of the rectangular conversion field which buries and puts a series by inverse transform Symmetry restrictions of the Fourier coefficient for being obtained had received restrictions. Namely, symmetry By the restrictions by it being a conjugate complex number, the Fourier coefficient of a position should embed an embedding series. It is substantially restricted to the half of the whole frequency coefficient, and a Frequency coefficient is spectral diffusion. It was difficult to lengthen diffuse series length.
[0016] 'synchronous subject:
Electronic Penetration The problems following in the conventional technique for a space synchronization and a time synchronization A point is rash.
[0017] It is It is a force, if it becomes credit mosquito Processing of time first by the method according to comprehensive search to perform search to all the amounts of synchronous displacement very much and it is realistic.
[0018] With the method of embedding the signal for matching synchronization, only the part of a matching synchronization signal is to a signal. The receiving amount of changes was made to increase and the quality of the signal was lowered as a whole. For example, place of an image It was connected with degrading the quality of an image in Synthesis. the signal for matching synchronization itself it embeds, and contributes as a noise ingredient to detection of information, and a detection function deteriorates -- possible There was also a sex. As for the characteristic signal for matching synchronization, it predicts and Ease itself is an attack. It may have been applicable and may have degraded the security of electronic Power.
[0019] The following subjects occurred especially about the time synchronization of the image.
[0020] Take a photograph with cameras displayed on a screen, TV, etc., such as an image power S video camera and a mobile phone. When carried out, since the reproductive frame rate and the frame rate of photography did not synchronize, the re-sampling with a subframe produced them, and the synchronization was much more difficult. Carrying When using processors of low performance, such as an electrification talk, the frame rate of photography is not stabilized. The timing of a sampling may shift minutely and this also synchronizes difficult. It had become a cause.
[0021] In the present invention, it was made in view of the above-mentioned point. therefore, tolerance can lengthen high Diffusion series length also to high compression or re-photography, and a time synchronous' space synchronization becomes unnecessary -- or. Time synchronization Electronic Penetration which can take - space synchronization easily Embedding art and electron It aims at providing watermark detection art.
Means for solving problem
[0022] In the electronic watermark embedding device with which the present invention has an embedding series creating means, an arrangement creating means, a modulation means, a memory measure, and an embedding pattern superposing means, N (N is two or more Integer) Embed to an input signal with the above dimension, carry out electronic With power of the information, and he is man. It is an electronic watermark embedding method embedded so that it may not be detected by consciousness, being the above-mentioned -- burying A lump series creating means generates an embedding series based on the above-mentioned embedding information. It stores in the memory measure of The 1, and the above-mentioned arrangement creating means is the above-mentioned embedding system of the memory measure of above-mentioned The 1. It is based on a sequence and is N. -- A one-dimensional pattern is generated, The above-mentioned modulation means is the above-mentioned N. -- One-dimensional pattern The embedding pattern of N dimension is generated by modulating a periodic signal according to the upper value, It stores in the 2nd memory measure and the above-mentioned embedding pattern superposing means is a memory measure of above-mentioned The 2. It is stored, the embedding pattern of and a The N dimension is acquired, and it is the embedding pattern. It is considered as the electronic watermark embedding method characterized by the embedded thing to superimpose on the above-mentioned input signal. Composed of can be carried out.
[0023] The present invention is an embedding series creating means, an arrangement creating means, a conversion method, a memory measure, an embedding pattern superposing means, and an inverse transform means, the electronic watermark embedding device which Have -- V, The, and N -- embedding to an input signal with the above (N is integer greater than or equal to 2) dimension -- information -- Electricity Electronic Penetration embedded so that Child permeability and may be carried out and it may not be detected by man's consciousness The method of embedding it is -- the above-mentioned embedding series creating means, It is based on the above-mentioned embedding information, and is an embedding system. A sequence is generated and it stores in the memory measure of The 1, The above-mentioned arrangement creating means is a memory measure of above-mentioned The 1. It is based on the above-mentioned embedding series stored, and is N. -- A one-dimensional pattern is generated, It stores in the memory measure of The 2, and the above-mentioned conversion method carries out rectangular conversion of the above-mentioned input signal, Changed signal It acquires and the above-mentioned embedding pattern superposing means is stored in the memory measure of above-mentioned The 2. The above-mentioned N -- A one-dimensional pattern is superimposed on some N--primary planes of the above-mentioned changed signal, inverse transform acquire a last signal, and the above-mentioned inverse transform means carries out rectangular inverse transform of the above-mentioned signal before inverse transform, and buries it -- Included Also constitute what it sees and acquires a settled signal for as an electronic watermark embedding method by which it is characterized. It can do.
[0024] Furthermore, the present invention is a demodulation means, a detection series extraction means, a correlation value calculating means, and a memory measure, Digital-watermarking detecting device Well which Have, The, and N Input with the above (N is integer greater than or equal to 2) dimension Electronic Penetration embedded so that it might not be beforehand detected by man's consciousness to a signal Detection It is an electronic watermark detecting method to carry out, The above-mentioned demodulation means is one dimension of the above-mentioned input signal. The ingredient of the predetermined periodic signal in a direction is measured, N -- It asks for a one-dimensional pattern and is the above-mentioned Inspection. An appearance series extraction means is the above-mentioned N. -- Value power of a one-dimensional pattern A detection series is searched for, memory measure Store and the above-mentioned correlation value calculating means buries with the above-mentioned detection series stored in the above-mentioned memory measure. digital watermarking currently embedded is detected based on the size of the correlation value of a lump series -- things can also be constituted as an electronic watermark detecting method by which it is characterized.
[0025] The present invention can also be constituted as a program which makes a device suitable for operation of the above-mentioned all directions method, and a computer perform the processing procedure of the describing [ above ] all directions method.
EFFECT OF THE INVENTION
[0026] Like [ according to the present invention / video irreversible numerals Y of high compression the image by which the image outputted to the display device was re-photoed, etc. ], Change greatly to the signal after digital-watermarking embedding When added, there is sufficient tolerance, quality degradation is suppressed and information length is long. The art which carries out electronic With power of the information and embeds it is realizable. the needlessness of matching synchronization -- if -- which can realize art of To being easy and detecting possible electronic Power of matching synchronization at high speed. That is, increase of processing time and embedding of a synchronized signal by synchronous processing Signal degradation to depend is prevented and tolerance and detection performance are high digital watermarking with little quality degradation at a high speed. It embeds and Z detection is attained. [ according to the present invention ]
Brief explanation of the drawings
[0027] [-- figure 1A] -- it is a flow chart which shows the outline of the electronic watermark embedding method in an embodiment of the invention.
[figure 1B] It is Frochia which shows the outline of the electronic watermark detecting method in an embodiment of the invention - A.
[figure 2] An electronic watermark embedding device and a digital-watermarking detecting device in an embodiment of the invention It is a figure showing outline composition.
[figure 3] N = it is an example of the embedding signal after the abnormal conditions at the time of 2.
* It is example 1 of composition of a 4A] cycle signal.
* It is example 2 of composition of a 4B] cycle signal.
* It is example 3 of composition of a 4C] cycle signal.
[figure 5A] It is Example 1 of the autocorrelation function of a periodic signal.
[figure 5B] It is Example 2 of the autocorrelation function of a periodic signal.
[figure 5C] It is Example 3 of the autocorrelation function of a periodic signal. * It is locus 1 on the complex plane of function hO according to a 6A] cycle signal.
* It is locus 2 on the complex plane of function hO according to a 6B] cycle signal.
* It is locus 3 on the complex plane of function hO according to a 6C] cycle signal.
[figure 7A] It is Example 1 of two periodic signals which intersect perpendicularly.
[figure 7B] It is Example 2 of two periodic signals which intersect perpendicularly.
* It is Example 1 of the signal after an 8A] synchronous gap.
* It is Example 2 of the signal after an 8B] synchronous gap.
[figure 9] It is an example of the locus of Q (X) by synchronous gap.
[figure 10] Electronic Penetration in the embodiment of The 1 of the present invention An embedding device and Electronic penetration It carries out and is an example of composition of a detecting device.
[figure 11] Flo of operation of the electronic Penetration embedding device in the embodiment of The 1 of the present invention It is one chart.
* It is an example of composition of the complex pattern generation part in the embodiment of The 1 of 12] present invention.
[figure 13] It is Frochia of processing of the complex pattern generation part in the embodiment of The 1 of the present invention - A.
* Off of detailed operation of the embedding series generation part in the embodiment of The 1 of 14] present invention It is a low chart.
[figure 15] It is an example of composition of the symbol in the embodiment of The 1 of the present invention.
* flow chart of operation of the complex arrangement generation part in the embodiment of The 1 of 16] present invention it is .
* It is an example of composition of the complex arrangement in the embodiment of The 1 of 17] present invention.
* It is an example of composition of the time modulation part in the embodiment of The 1 of 18] present invention.
* It is a flow chart of operation of the time modulation part in the embodiment of The 1 of 19] present invention, and is Ah. To.
[figure 20] Repeat the embedding pattern in the embodiment of The 1 of the present invention in the direction of time. It is an example to superimpose.
[figure 21] Carry out tiling in all directions [ of the embedding pattern in the embodiment of The 1 of the present invention ]. It is an example to superimpose. [figure 22] example which expands and superimposes the embedding pattern in the embodiment of The 1 of the present invention it is .
* It is Frochia of operation of the digital-watermarking detecting device in the embodiment of The 1 of 23] present invention - A.
* Compute the characteristic quantity of the embedding finishing signal in the embodiment of The 1 of 24] present invention. It is an example to extract.
* It is an example of composition of the time demodulation section in the embodiment of The 1 of 25] present invention.
[figure 26] It is a flow chart of operation of the time demodulation section in the embodiment of The 1 of the present invention, and is Ah. To.
* example of composition using the difference' differentiation of the time demodulation section in the embodiment of The 1 of 27] present invention it is .
* It is an example of composition of the detection information extraction part in the embodiment of The 1 of 28] present invention.
* flow chart of operation of the detection information extraction part in the embodiment of The 1 of 29] present invention it is .
* It is a flow chart of detailed operation of the detection series extraction part in the embodiment of The 1 of 30] present invention.
* It is an example of composition of the time modulation part in the embodiment of The 2 of 31] present invention.
* It is a flow chart of operation of the time modulation part in the embodiment of The 2 of 32] present invention, and is Ah. To.
* It is an example of composition of the time demodulation section in the embodiment of The 2 of 33] present invention.
[figure 34] It is a flow chart of operation of the time demodulation section in the embodiment of The 2 of the present invention, and is Ah. To.
* example of composition using the difference' differentiation of the time demodulation section in the embodiment of The 2 of 35] present invention it is .
* It is an example of composition of the complex pattern generation part in the embodiment of The 3 of 36] present invention. * It is Frochia of operation of the complex pattern generation part in the embodiment of The 3 of 37] present invention - A.
* Complex arrangement in the complex arrangement generation part in the embodiment of The 3 of the 38A] present invention It is Example 1 of the element range.
[figure 38B] It is Example 2 of the element range of the above-mentioned complex arrangement.
[figure 38C] It is Example 3 of the element range of the above-mentioned complex arrangement.
[figure 38D] It is Example 4 of the element range of the above-mentioned complex arrangement.
[figure 38E] It is Example 5 of the element range of the above-mentioned complex arrangement.
[figure 38F] It is Example 6 of the element range of the above-mentioned complex arrangement.
[figure 39A] It is Example 7 of the element range of the above-mentioned complex arrangement.
[figure 39B] It is Example 8 of the element range of the above-mentioned complex arrangement.
[figure 39C] It is Example 9 of the element range of the above-mentioned complex arrangement.
[figure 39D] It is Example 10 of the element range of the above-mentioned complex arrangement.
[figure 39E] It is Example 11 of the element range of the above-mentioned complex arrangement.
[figure 39F] It is Example 12 of the element range of the above-mentioned complex arrangement.
* It is an example of composition of the detection information extraction part in the embodiment of The 3 of 40] present invention.
* flow chart of operation of the detection information extraction part in the embodiment of The 3 of 41] present invention it is .
* It is an example of composition of the detection information extraction part in the embodiment of The 4 of 42] present invention.
* flow chart of operation of the detection information extraction part in the embodiment of The 4 of 43] present invention it is .
* It was which explains an example of a detecting method for the bit value in the embodiment of The 4 of 44] present invention Because -- it is a figure.
* It is an example of composition of the digital-watermarking detecting device in the embodiment of The 5 of 45] present invention. * It is Frochia of operation of the digital-watermarking detecting device in the embodiment of The 5 of 46] present invention - A.
* It is an example of composition of the synchronous primary detecting element in the embodiment of The 5 of 47] present invention.
* It is a flow chart of operation of the synchronous primary detecting element in the embodiment of The 5 of 48] present invention, and is Ah. To.
* It is an example of composition of the detection information extraction part in the embodiment of The 5 of 49] present invention.
* They are other examples of composition of the detection information extraction part in the embodiment of The 5 of 50] present invention. * It is an example of composition of the time modulation part in the embodiment of The 6 of 51] present invention.
[figure 52] The electronic watermark embedding device and Electronic penetration in an embodiment of The 7 of the present invention It carries out and is an example of composition of a detecting device.
[figure 53] Flo of operation of the electronic Penetration embedding device in the embodiment of The 7 of the present invention It is one chart.
* It is an example of composition of the complex pattern generation part in the embodiment of The 7 of 54] present invention.
[figure 55] It is Frochia of operation of the complex pattern generation part in the embodiment of The 7 of the present invention - A.
* It is Frochia of operation of the digital-watermarking detecting device in the embodiment of The 7 of 56] present invention - A.
* It is an example of composition of the detection information extraction part in the embodiment of The 7 of 57] present invention.
* It is an example of combination of the amount of synchronous displacement in the embodiment of The 7 of 58] present invention.
[figure 59] Flo of operation of the electronic Penetration embedding device in the embodiment of The 8 of the present invention It is one chart.
* It is an example of composition of the complex pattern generation part in the embodiment of The 8 of 60] present invention.
[figure 61] It is Frochia of operation of the complex pattern generation part in the embodiment of The 8 of the present invention - A.
* A plurality of information which can be set to the embodiment of The 8 of 62] present invention is continuously filled up with time sharing, and it is Included. It is an example of Mu.
* It is an example of composition of the digital-watermarking detecting device in the embodiment of The 8 of 63] present invention. * It is Frochia of operation of the digital-watermarking detecting device in the embodiment of The 8 of 64] present invention - A.
[figure 65] Position power from which the synchronization in the embodiment of The 8 of the present invention shifted Synchronous pattern It is an example of detection.
* It is an example of composition of the detection information extraction part in the embodiment of The 8 of 66] present invention.
* flow chart of operation of the detection information extraction part in the embodiment of The 8 of 67] present invention it is .
[figure 68] The electronic watermark embedding device and Electronic penetration in an embodiment of The 9 of the present invention It carries out and is an example of composition of a detecting device.
[figure 69] Flo of operation of the electronic Penetration embedding device in the embodiment of The 9 of the present invention It is one chart.
Explanations of letters or numerals
102 The 1st memory measure
103 Memory Measure of The 2
100 Electronic Watermark Embedding Device
110 Complex Pattern Generation Part
111 Embedding Series Creating Means, Embedding Series Generation Part
112 Arrangement Creating Means, Complex Arrangement Generation Part
113 N -- One-dimensional Inverse Fourier Transform Part
114 Embedding Information Dividing Part
115 Synchronous Series Generation Part
116 Complex Arrangement Generation Part
117 Embedding Series Generation Part
120 Arrangement Generation Part
130 Modulation Means, Time Modulation Part
131 Periodic Signal Generating Part
132 Modulation Part
133 Adder Unit
134 One-dimensional Inverse Fourier Transform Part
136 Modulation Part
140 Embedding Pattern Superposing Means, Embedding Pattern Superposed Part
150 Storage Part of The 1
160 Storage Part of The 2
200 Digital-Watermarking Detecting Device
202 Memory Measure
210 Demodulation Means and Time Demodulation Section 211 Periodic Signal Generating Part
212 Demodulation section
213 Complex Pattern Formation Part
214 One-dimensional Fourier Transform Part
215 Signal Differentiation Part
216 One-dimensional Fourier Transform Part
220 Detection Information Extraction Part
221 Detection Series Extraction Means, Detection Series Extraction Part
222 A correlation value calculating means, a correlation value calculating part
223 Maximum judgment part
224 Detection Information Reconstruction Section
225 N -- One-dimensional Fourier Transform Part
226 Complex Correlation Value Calculating Part
227 Absolute Value Calculation Part
250 Pattern Memory Part
300 Digital-Watermarking Detecting Device
310 Time Demodulation Section
320 Synchronous Primary Detecting Element
321 Complex Detection Series Extraction Part
322 Complex Correlation Value Calculating Part
323 Absolute Value Calculation Part
324 Synchronous Detection Maximum Judgment Part
325 Phase Calculation Part
330 Detection Information Extraction Part
331 Detection Series Extraction Part
332 Correlation Value Calculating Part
333 Maximum Judgment Part
334 Detection Information Reconstruction Section 340 Pattern Memory Part
500 Electronic Watermark Embedding Device
510 Complex pattern generation part
511 Embedding series generation part
512 Complex arrangement generation part
520 Time Modulation Part
530 Embedding Pattern Superposed Part
600 Digital-watermarking detecting device
610 Time demodulation section
620 Synchronous Primary Detecting Element
630 Detection Information Extraction Part
631 Detection Series Extraction Part
632 Correlation Value Calculating Part
633 Maximum Judgment Part
634 Detection Information Reconstruction Section
700 Digital-watermarking detecting device
710 Embedding finishing signal dividing part
720 Synchronous time demodulation section
730 Synchronous Primary Detecting Element
740 Synchronized Signal Dividing Part
750 Time demodulation section
760 Detection Information Extraction Part
761 Detection Series Extraction Part
762 Correlation Value Calculating Part
763 Maximum Judgment Part
764 Detection Information Reconstruction Section
765 Detection information connecting part
800 Electronic Watermark Embedding Device 810 Complex Pattern Generation Part
820 Embedding pattern superposed part
830 The signal conversion section before embedding
840 Embedding finishing signal inverse transforming part
850 The 1st storage part
904 Middle Complex Pattern
911 Embedding information
912 Signal before Embedding
913 Embedding Series
914 Detection Information
917 Synchronous Series
921 Embedding Complex Pattern
922 Embedding pattern
923 Embedding finishing signal
961 Detection Complex Pattern
1113 Detection series
1114 Correlation Value
1115 Detection Complex Arrangement
1116 Complex Correlation Value
1117 Absolute Value
1118 Detection Complex Number Series
1501 Detection complex pattern
1502 The amount of synchronous displacement
1511 Detection Complex Series
1512 Complex Correlation Value
1513 Absolute Value
1521 Detection Series
1522 Correlation Value 3111 Embedding Information
3112 Signal before Embedding
3113 Embedding Finishing Signal
3114 Detection Information
3121 Embedding Complex Pattern
3122 Embedding Pattern
3161 Detection Complex Pattern
3162 Amount of Synchronous Displacement
3213 Embedding Series
3313 Detection Series
3314 Correlation Value
3613 Detection Series
3614 Correlation Value
3615 Partial Detection Information
3812 Detection Information
3813 Synchronous Complex Pattern
3814 Amount of Synchronous Displacement
3815 Detection Complex Pattern
3816 Partial Embedding Finishing Signal
3817 Synchronized Partial Signal
4021 Embedding Complex Pattern
4022 Signal before Changed Embedding
4023 Front [ Inverse Transform ] Embedding Finishing Signal
The best form for inventing
First, the outline of an embodiment of the invention is explained. Figure 1A is an embodiment of the invention. Electronic Penetration which can be set It is a flow chart which shows the outline of the method of embedding.
The above-mentioned electronic watermark embedding methods are an embedding series creating means, an arrangement creating means, and abnormal conditions. A means, a memory measure, and embedding pattern superposing means, electronic watermark embedding device which Have , The, and N -- embedding to an input signal with the above (N is integer greater than or equal to 2) dimension -- Sentiment carrying out in Electronic penetration embedded so that electronic With power of the news may be carried out and it may not be detected by man's consciousness, and burying -- Included It is the method of seeing. An embedding series creating means buries this method based on embedding information. Embedding series generation step (Step 1) which generates a lump series and it stores in the memory measure of The 1, An arrangement creating means is based on the embedding series of the memory measure of The 1, and it is N. -- One-dimensional putter Arrangement generation step (Step 2) which generates A, A modulation means is N. -- On a one-dimensional pattern Abnormal-conditions step (Step 3) which generates the embedding pattern of N dimension and it stores in the memory measure of The 2 by modulating a periodic signal according to a value, An embedding pattern superposing means is stored in the 2nd memory measure, and acquires the embedding pattern of and a To N dimension, and it is the embedding. It has an embedding pattern superposition step (Step 4) which superimposes a pattern on an input signal. ing.
[0030] a figure -- 1B is a flow chart which shows the outline of the electronic watermark detecting method in an embodiment of the invention. This electronic watermark detecting method is a demodulation means, a detection series extraction means, and correlation. In the digital-watermarking detecting device which has a value calculating means and a memory measure, Dimension more than N Electronic penetration embedded so that it might not be beforehand detected by man's consciousness to the input signal which it has It is an electronic watermark detecting method which detects power.
[0031] Predetermined periodic signal [ in / method / this gentleman / in a demodulation means / the one direction of a dimension of an input signal ] An ingredient is measured and it is N. -- Recovery step (Step 11) which asks for a one-dimensional pattern, Detection system A sequence extraction means is N. -- Value power of a one-dimensional pattern A detection series is searched for and it stores in a memory measure. Detection series extraction step (Step 12), The correlation value calculating means was stored in the memory measure. Electronic penetration which embeds with a detection series and is embedded based on the size of the correlation value of a series It has a correlation value calculation step (Step 13) which detects To.
[0032] a figure -- 2 is a figure showing the outline composition of an electronic watermark embedding device and a digital-watermarking detecting device in an embodiment of the invention.
[0033] An electronic Penetration embedding device in this embodiment, N (N is an integer greater than or equal to 2) Embed to an input signal with the above dimension, and perceive in man's consciousness by making information into digital watermarking. It is electronic watermark embedding device 100 embedded so that it may not be carried out, It is Based to embedding information. Embedding series creating means 1 which comes, and generates an embedding series and it stores in memory measure 102 of The 1 11, It is based on the embedding series of memory measure 102 of The 1, and is N. -- A one-dimensional pattern is generated. Arrangement creating means 112, N -- A periodic signal is modulated according to the value on a one-dimensional pattern. Modulation means 1 which generates the embedding pattern of N dimension and it stores in memory measure 103 of The 2 30, the embedding pattern of N dimension stored in memory measure 103 of The 2 is acquired -- embedding pattern superposing means 140 which superimposes the embedding pattern on an input signal -- an owner -- It carries out.
[0034] A digital-watermarking detecting device of this embodiment, Electronic Penetration embedded so that it might not be beforehand detected by man's consciousness to an input signal with the dimension more than N Electron to detect It is watermark detecting device 200, Predetermined periodic signal in the one direction of a dimension of an input signal An ingredient is measured and it is N. -- Demodulation means 210 which asks for a one-dimensional pattern, N -- One-dimensional pattern Detection series extraction means 221 which Value power also searches for a detection series and stores in memory measure 202, account it embeds with the detection series stored in Means 250, and is based on the size of the correlation value of a series -- it has correlation value calculating means 222 which detects digital watermarking currently embedded.
[0035] Hereinafter, describe an embodiment of the invention in detail with a drawing.
[0036] Below, after the basic concept used as the principle of the present invention and its example are shown first, describe the embodiment of a concrete electronic watermark embedding device, a digital-watermarking detecting device, and its method. It carries out.
[0037] [Basic concept]
First, the basic concept used as the principle of this embodiment is shown below.
[0038] Electronic Penetration model by (1) complex correlation:
Correlation use type digital watermarking using a complex number sequence is shown below.
[0039] As the concept that the portion made into the complex number in this specification of still the following is the same It cannot be overemphasized that it can transpose to two-dimensional Bettoru.
[0040] Define signal sequence i for embedding embedding, and digital-watermarking series w as follows.
[0041] i= {i, i, - - -, and i} ≡C<sup>L</sup>(1)
1 2 L
w= {w, w, - - -, and w} £ C<sup>L</sup>(2)
1 2 L
however, C is a set of the whole complex number -- w is taken as the pseudorandom-numbers sequence of average 0.
[0042] Embed w at i and obtain embedding finishing signal sequence i'. [0043] i' =i+w (3)
In detection i', the phase shifted only △ theta by synchronous gap. gamma is considered.
[0044] [One number]
(4)
Here, j is an imaginary unit.
[0045] Calculate the correlation value of gamma and w like a following formula, and detect electronic Power.
[0046] p =i"-w* (5)
"here -- w*" is a sequence of numbers" which takes the conjugate complex number of each element of w -- a sequence of numbers -- vector It is an inner product operation when it sees. It is complex correlation about the correlation computation of such complex number sequences. It will call.
[0047] [Two number]
(6)
*w
<img file="WO2007102403A1_D0001.tif" />
Here, W* expresses the conjugate complex number of W.
k k
If i and w are independent and L is large enough, the expected value of ∑i w* is 0.
k k
[0048] [Three number]<img file="WO2007102403A1_D0002.tif" />
It follows,
* An absolute value is not concerned with amount of phase gaps △ theta by calculating the embedding value used as the maximum. It becomes detectable. That is, it becomes detectable without matching synchronization.
[0049] Calculation of amount of phase gaps △ theta is attained by searching for the angle of deviation of -.
[0050] As w, the two following kinds of how to choose can be considered here.
k
[0051] 1, for example, {+1, and -- case; where it is chosen only out of the real number as it said that one value of 1} was taken
2) For example,
[0052] [Four number]<img file="WO2007102403A1_D0003.tif" />
of! -- if it shifts and that value is taken! /and Ivy -- case; where it is chosen so that it may be spread and distributed over complex number space like
In above 1, the phase of pattern ingredient w of Power is All together on embedding finishing signal sequence i'. The phase will be spread when it is Am and 2. It is ∑iw* even if it is one of cases.
k k
Expected value is I w, although it is set to o and detection of electronic Power is attained.
k Make distribution of I equal. The direction of 2 is I ∑i w* the case where 1 and 2 are compared under the conditions which arranged the energy of the digital-watermarking ingredient.
k k I
Dispersion of becomes small. Noise ingredient which embeds this at the value of p and target signal sequence i gives It means that influence becomes small and more reliable detection is attained.
[0053] Time synchronous unnecessary digital watermarking by the direction single frequency embedding of (2) time: As an example of digital-watermarking embedding from which a phase shifts by synchronous gap only in △ theta, it is time. The example for performing embedding to a direction single frequency is shown.
[0054] The example which specifically modulates and embeds the complex number pattern which comprised a direction of space of the image at the single frequency of the direction of time, for example is shown.
[0055] It is N to the signal sequence of an embedding N dimension. -- One-dimensional complex number pattern P(x) is embedded. however
X is N. -- It is a vector showing the position in one-dimensional space,
chi= (chi, X, ..., x) (8)
1 2 N-l
It carries out.
[0056] Consider the Nth dimension to be a time-axis and prepare periodic function of complex variable f (t) of the direction of a time-axis. For example,
[0057] [Five number]
f e (9)
= cos omega*+j sin omega*
As for j, an imaginary unit and omega are f here. It is the angular velocity showing the cycle of (t).
[0058] the real part of P (X), and an imaginary part -- respectively -- f by the real part of (t), and the imaginary part, AM abnormal conditions were carried out and it compounded -- N dimension real number signal W (x, t) is acquired.
[0059] Garden
W(.x, *) = [Ws^cos omega *+3[rho (]) sin omega* (1 0)
= omega><< *omega]
<img file="WO2007102403A1_D0004.tif" />
However, f* (t) is a function of complex variable which takes the conjugate complex number of t.
[0060] the time of N= 2 -- the example of W (x, t) -- a figure -- it is shown in 3. A thick line expresses the value of w (x, t), is expressed by the periodic signal of the direction of t axis where the phase shifted every position X here, and they are ! and To.
[0061] By such abnormal conditions, it is f at the angle of deviation and the absolute value of the complex number value of P(x). Strange in the phase and amplitude of (t) A tone is carried out and it is To make * This.
[0062] Embed W (x, t), embed to target signal I (X, t), and acquire an embedding finishing signal (X, t).
[0063] I' (x, t) =l (x, t) +W(x, t) (11)
Signal from which detection (chi, t) shifted only At by synchronous gap gamma (chi, t) is considered.
[0064] [Seven number]
J "(x, t) = l '(x + M) (1 2)<img file="WO2007102403A1_D0005.tif" />
for example, re-photography according to a video camera when a picture signal etc. are considered and an analog -- conversion also when When is performed, even if the positions on space usually differ -- synchronous displacement of the direction of time Quantity becomes fixed. (When a direction of time to differ for every spatial position as attack on digital watermarking said term gap is given) Image quality deteriorates remarkably and the value as image contents is disadvantage. From this which is not realized as an attack since it is divided, as mentioned above, the amount At of synchronous displacement of the direction of time (the direction of t) cannot call at position X, but can be considered to be fixed.
[0065] For example, when an image signal was considered, even if only the parallel translation of the picture was added and the image signal differed in position X of the transverse direction in To go case, for example, it was added to the lengthwise direction. The size of parallel translation becomes fixed. Also in this case, it is the synchronization of a lengthwise direction (the direction of t) as mentioned above. The amount At of displacement cannot call at position X of a transverse direction, but can be considered to be fixed.
[0066] It was given now by section 0*t*T (T is an integral multiple of the cycle of f (t)). It is f to gamma (chi, t) (t). The following integration is calculated in order to get over.
[0067] [Eight number] Qix= I"(x, T) f*(T) dr (1 3)
- j em
=
12 sigma
<img file="WO2007102403A1_D0006.tif" />Above
T<sub>e</sub>)
= N{x and At+ ------ P(x) -- here, it is nu(chi, alpha*) = * e (x, r+A/).
Jo
However, the following relation was used.
[0068] [Nine number]
<img file="WO2007102403A1_D0007.tif" />
^ . : +b - *tau
2 A A*f
(a + jb)
2
It comes out and is from a certain thing.
[0069] [Ten number]
-jw(,T J<sub>c</sub>- ji>>r The<sub>=</sub>
(1 5) In addition at an example here, it is a signal. Although the case where gamma (chi, t) was obtained as a continuation signal was explained to the example, also when signal I" (x, t) is obtained as a discrete signal, the same processing is possible by Sum of products calculation instead of integration.
[0070] namely
[0071] [11 number]
(16)<img file="WO2007102403A1_D0008.tif" />
omegadelta*
(a + jb)
It comes out and is from a certain thing,<img file="WO2007102403A1_D0009.tif" />
gamma-1
2 A (18)
tau-=0
<img file="WO2007102403A1_D0010.tif" />
Te
= nu however A*c A= y, /0c r+A ^_^<sup>r</sup>It comes out.
r=0
Watermark series
w -- {-- w and .. * and w}
1
Linear transformation Trans with P(x) and inverse transform Trans1<sup>1</sup>the time of Definition(ing) -- digital watermarking embedding is faced -- P(x) = Trans (w) (19)
It comes out, P(x) is generated and digital-watermarking detection is faced,
w' = Trans" '(QCx)) (20)
It is if it carries out,
[A 12 number]
w' - n/ w (2 1)
2
Can be obtained. However, n=Trans (N (x, At)) <sub>G</sub>
[0073] As an example of linear transformation Trans, it is only N about each value of w, for example. -- It may be conversion of only To be in order on one-dimensional space, and is N about each value of w. -- After arranging in order on one-dimensional space It may be the conversion which performs rectangular conversion of inverse Fourier transform etc.
[0074] It is applying the argument on the foregoing paragraph and taking complex correlation,
[0075] [13 number]
rho = omega - co
= (<<+ (2 2))
* fe∑<sup>2</sup>
Detection of power electronic Power and calculation of the amount At of synchronous displacement are possible.
[0076] (3) -- difference -- detection: by correlation (differentiation)
it sets to the above-mentioned detection method -- f (t) is used and it may be made to recover from the difference of I" (x, t) or , and differentiation instead of the calculation which asks for Q(x)
[0077] Namely, when a signal is discrete, for example,
<img file="WO2007102403A1_D0011.tif" />
It carries out, [0078] [14 number]
gamma-1
r=0
rho (chi) (2 3)<img file="WO2007102403A1_D0012.tif" />
= O (x< delta*) - nu<sub>0</sub>(chi. At)+~-~rho (chi)
Tau-iota tau
However, lambda^omicron (chi, delta*) = J iota(chi< r+M) e -<sup>}03</sup>^ delta*= > ./(chi and gamma+delta* 1 [ ^ ^ ] -- it comes out.)
tau=0 r=l
It is here,
[0079] [15 number]
e}<sup>&</sup>- \
Since it is Is known, the same detection is possible.
[0080] When a signal is continuous for example,
[0081] [16 number]
Differentiation ^^
By the same calculation as Was used, the same detection is possible.
[0082] For example, when a picture signal etc. are considered, it is known that inter-frame correlation of a signal is large. If inter-frame difference is calculated as mentioned above from inter-frame correlation being high so that the time interval of two frames is short, more than the offset part by a periodic signal -- original picture ingredient being canceled greatly -- the absolute value of the paragraph of the above-mentioned N (X, At) -N (X, At) -- very much -- small -- in a fence, the energy of electronic Power increases relatively -- detection -- Shabu -- Is called advantage It is.
[0083] (4) They are business and time synchronous unnecessary digital watermarking of a It was case about arbitrary periodic signals:
Next, using functions f (t) arbitrary as a periodic signal, the phase of this is changed and it is digital watermarking. The case where it embeds is considered. Here, it is f. (t) considers a real function.
[0084] the absolute value of embedding P (X), and the angle of deviation -- f the amplitude of (t) and a phase were modulated -- N dimension real number signal
W (x, t) is obtained.
[0085] For example,
W(x, t)= I P(x) I f(t+s(x)) (24)
It is here,
[0086] [17 number]
six) = i~ Arg[p(x)]
2 pi
T is f. Suppose that it is a cycle of (t).
[0087] Embed W (x, t), embed to target signal I (X, t), and acquire an embedding finishing signal (X, t).
[0088] I'(x, t) =l(x, t) +W(x, t) (25)
Signal from which detection (chi, t) shifted only At by synchronous gap gamma (chi, t) is considered.
[0089] I"(x, t)=I'(x, t+At) (26)
=1 (x, t+At)+W(x, t + At)
=l(x, t+At) + I P(x) I f(t+At+s(x))
now, it was given by section 0*t*T as opposed to gamma (chi, t) -- phase of f (t) and f (t) pizeta4 displacement The following integration for getting over by f (t- pi / 4) carried out is calculated.
[0090] [18 number] (2 7) <img file="WO2007102403A1_D0013.tif" />
here -- N, x, ^= f/C, and The +delta*- /()+# (r -- suppose that it is --.)
This integration calculation is the calculation same with asking for the autocorrelation of f(), and is f. Self of (t) If a correlation function is set to g (t).
[0091] [19 number]
(2 8) /(r+Ai+s(x)) f(T) dr = g(At+six) zeta/(r+delta*+six) f r - +six+ (2 9)
<img file="WO2007102403A1_D0014.tif" />
It can express.
[0092] Now,
[0093] [20 number]
h(x, delta*) = g(M+six)+jg^At+six+^j (30)
It is if it carries out,
[0094] [21 number]<img file="WO2007102403A1_D0015.tif" />
becoming power Since detection of the same electronic Power as the example using the sine wave of the future above-mentioned can be performed To -- phase of the signal with which angle of deviation Arg [h] (x, At) of h (x, At) was embedded
[0095] [22 number]<img file="WO2007102403A1_D0016.tif" />
It might be alike and may have taken the near value. namely
[0096] [23 number]
Arg [h{x and delta*] = ---- (delta*+six) +alphaphi w 2
It may be function h that it comes out and △ phi takes a sufficiently small value.
[0097] At this time, the product operation in complex correlation is expressed as follows,
[0098] [24 number]
<img file="WO2007102403A1_D0017.tif" />
It will ask for total of this as a complex correlation value, and is synchronous displacement from the angle of deviation of a complex correlation value. It can ask in the range with error according to quantity At and △ phi.
[0099] Correct, [0100] [25 number]<img file="WO2007102403A1_D0018.tif" />
It came out and used a certain thing. In order to understand easily, the example which excludes linear transformation Trans and performs the correlation operation in direct P (x) was shown.
[0101] When a periodic signal is the above argument a sine wave, while being set to △ phi =0, the case of periodic signals other than a sine wave, it is set to !=0 and means that an error arises in measurement of the amount At of synchronous displacement. ing.
[0102] Example of (5) cycle signals:
A periodic function with the following feature should just be used for a periodic signal.
[0103] The result with which it integrated by 1 1 cycle is set to 0.
[0104] It is a peak with two sharp autocorrelation functions Do not have !
[0105] the above-mentioned conditions of 1 -- for example
[0106] [26 number]<img file="WO2007102403A1_D0019.tif" />
It may carry out. Here tau is a cycle of a periodic function.
[0107] The conditions of above 2 may use the conditions "the value of the second degree differentiation near the peak of an autocorrelation function is not in agreement with the numerals of the peak."
[0108] The example of a periodic signal is shown in figure 4 A*C. It is a periodic signal (figure 4A) of the figure, the (a) sine wave (Drawing 4 B), the (b) triangular wave (figure 4C), and (c) rectangle wave, and is an example of 0*t To T (T is a cycle of a signal) about each signal. It is as follows when expressed with a formula in an enclosure. The above which drops off in what is limited to these examples Also until right and a thing say using arbitrary periodic signals with the feature /!
[0109] [27 number] (a) y = as mt
- . 0<sub>*f <</sub>
(b) y =
4 alpha "
<img file="WO2007102403A1_D0020.tif" />
The conditions of 1) are the results of the integration at the time of detection. The direct-current ingredient of the direction of t of gamma (chi, t) is cancellable. Things are expressed.
[0110] Explain the meaning of the conditions of 2 below.
[0111] The autocorrelation function of each periodic signal of figure 4 A*C is shown in figure 5 A*C. In the case of and each periodic signal, the locus to which function-of-complex-variable h() moves on a complex plane is shown at figure 6 A*C.
[0112] (Figure 5A and a figure 6A) In the example of a sine wave, the autocorrelation function is changing smoothly and the locus of h () serves as a circle. It is set to this and =0 and the amount At of synchronous displacement is calculated in high accuracy. It means.
[0113] (Figure 5B and a figure 6B) In the example of a triangular wave, Autocorrelation Seki which the autocorrelation function is changing smoothly after all, the value of the second degree differentiation near the peak differs from the numerals of the peak (the value of the second degree differentiation near [ positive ] the peak has negative numerals, and numerals are not in agreement), and a sharp peak does not have It is a number. a result -- M -- a locus becomes almost equal to a circle, delta phi serves as a sufficiently small value, and it means and that the amount At of synchronous displacement is too calculated in high and accuracy.
[0114] (Figure 5C and a figure 6C) In the example of a rectangle wave, Autocorrelation Seki which the value of the second degree differentiation near the peak differs from the numerals of the peak (numerals are Value power^ of the second degree differentiation near [ positive ] the peak, and numerals are not in agreement), and a too sharp peak does not have although an autocorrelation function cannot be said to be Sliding force It is a number. a result -- M -- although a locus is not a circle, it serves as a lozenge, and delta phi is comparatively small It becomes a value and means that the amount At of synchronous displacement is calculated in a certain amount of accuracy.
[0115] Generally it is Satisfied about the above-mentioned conditions, such as a triangular wave and a rectangle wave, instead of the direction which calculates the value of a rectangle wave or a triangular wave being a sine wave since it can process at high speed rather than calculating the value of a sine wave. It is using a periodic function besides To, High-speed [ as a whole ] at the sacrifice of the error of some of the amount of synchronous displacement are . Digital-watermarking detection is attained. For example, Digital watermark inspection in personal digital assistants, such as a mobile phone Extensive, when performing digital-watermarking detection in the very limited calculation resource like appearance It is effective, especially when performing digital-watermarking detection and high-speed processing is required.
[0116] Time synchronous unnecessary digital watermarking at the time of using two periodic functions which carry out (6) rectangular crosses:
Next, two periodic functions f which intersect perpendicularly with the fundamental frequency same as a periodic signal (t) and f
1 Consider the case where 2 t is used.
[0117] In this case, time synchronous unnecessary digital watermarking by the above-mentioned direction single frequency embedding of time Periodic function of complex variable f in explanation It is f about (t). (t) =f (t) +jf It is equal to being referred to as (t).
1 2
[0118] for example, f being referred to as (t) -- the rectangle wave of cycle 4, and f considering the triangular wave of cycle 4 as (t) -- respectively
1 2
Drawing 7A -- what is defined as a signal like B is considered.
[0119] this time -- f (t) and f both (t) has fundamental frequency 1Z4 -- integration for one cycle
1 2
[0120] [28 number]
I -, (i)<sub>2</sub>(/Wi - (3 5)
It becomes and lies at right angles. However, T is a signal cycle and it is T= 4 in this example.
[0121] In this case, W which corresponds, for example when it is certain 0 As opposed to chi = chi (x) rho (chi) = 1 +
0 0 j 0
t -- a figure -- f of 7 alpha It becomes the same as that of (t).
[0122] if it describes except for an original picture ingredient when there is no synchronous gap -- a figure -- signal G of 8A (t) -- the time of detection It is obtained. This and f (t) f Result of correlation computation with (t),
1 2
[0123] [29 number]<img file="WO2007102403A1_D0021.tif" />
■ G<sub>2</sub>(t)f<sub>2</sub>dt = it is set to 0 (3 7) and is Q (x) =4. + It is set to 0.
o j
[0124] if it describes except for an original picture ingredient when 90 degrees apart [ in phase ] by synchronous gap -- a figure -- a signal G like 8B (t) is obtained at the time of detection. This and f (t) and f Result of correlation computation with (t)
2 1 2
[0125] [30 number]
<img file="WO2007102403A1_D0022.tif" />
Next door,
[0126] [31 number]
4
rho<sub>0</sub>= omicron+ A
It becomes.
[0127] if it thinks the same way -- the size of a synchronous gap -- Q (x) -- a figure -- the complex plane top of 9 -- locus 1
0
Forget change will be carried out.
[0128] It is P (x) =0+ to x=x which exists, for example. W (x, t) which corresponds when it is 1 is f of Drawing 7B. Becoming the same as that of (t), Q (x) is by the size of a synchronous gap in the same consideration,
2 One Forget conversion of the locus 2 will be carried out for the complex plane top of Drawing 9.
[0129] By calculation of the complex correlation value mentioned above, and the same calculation, the angle of deviation of Q is side of a complex correlation value. Sumo can also calculate the amount At of synchronous displacement in the range of the error which becomes settled in calculation.
[0130] Two periodic functions f (t) which intersect perpendicularly with the fundamental frequency same as a periodic signal in this way
f This which calculates the amount of synchronous displacement as a whole at the sacrifice of some error also when (t) is used
2
But becomes possible.
[0131] [Embodiment of The 1]
An electronic watermark embedding device and an electron The composition of a watermark detecting device is shown. [ in / in Drawing 10 / the embodiment of The 1 of the present invention ]
[0132] <Electronic watermark embedding device>
First, an electronic watermark embedding device is explained.
[0133] Electronic watermark embedding device 100, comprising complex pattern generation part 110, time modulation part 130, embedding pattern superposed part 140, storage part 150 of The 1, and storage part 160 of The 2 -- Fill -- inputting lump information 911 and signal 912 before embedding -- embedding finishing signal 923 -- Output.
[0134] Below, explain operation of electronic watermark embedding device 100.
[0135] a figure -- 11 is a flow chart of operation of the electronic Penetration embedding device in the embodiment of The 1 of the present invention. Embedding place of digital watermarking by electronic watermark embedding device 100 Reason is carried out in the following procedures.
[0136] In step 100 complex pattern generation part 110, embed based on inputted embedding information 911, generate complex pattern 921, and store in storage part 150 of The 1, such as a memory. It carries out.
[0137] N which, as for embedding complex pattern 921, complex number power also comprises -- It is a one-dimensional pattern, the contents of embedding information are expressed, and they are and To. About is mentioned below in Drawing 12 in the details of operation of complex pattern generation part 110.
[0138] As shown in how to, embed others to the complex number mentioned below in addition, depending on processing of time modulation part 130, it is N. -- It is also good to constitute a one-dimensional pattern as a pattern of a real value. It is ability.
[0139] Step 110 In time modulation part 130, it is generated in complex pattern generation part 110, embeds based on embedding complex pattern 921 stored in the 1st storage part 150, and is The. Turn 922 is generated and it stores in storage part 160 of The 2, such as a memory. embedding pattern 92 2 modulates the direction of a time-axis to embedding complex pattern 921 -- a real value -- composition -- It generates as a To be N dimension pattern.
[0140] Mention the details of operation of time modulation part 130 below.
[0141] In step 120 embedding pattern superposed part 140, It was generated by time modulation part 130, embedding pattern 922 stored in storage part 160 of The 2 was inputted, and it buries. It superimposes on signal 912 before a lump, and outputs embedding finishing signal 923.
[0142] Mention the details of operation of embedding pattern superposed part 140 below.
[0143] Digital watermark embedding device-complex pattern generation part >
Drawing 12 is Show about the example of composition of the complex pattern generation part in the embodiment of The 1 of the present invention. To.
[0144] Complex pattern generation part 110a comprises embedding series generation part 111 and complex arrangement generation part 112, inputs embedding information 911, and outputs embedding complex pattern 921.
[0145] Generation processing of the embedding complex pattern by complex pattern generation part 110a is carried out in the following procedures.
[0146] a figure -- 13 is a flow chart of processing of the complex pattern generation part in the embodiment of The 1 of the present invention.
[0147] Generate embedding series 913 which is a numerical sequence showing embedding information in step 101 embedding series generation part 111 based on inputted embedding information 911. One! and The mention below in the details of operation of embedding series generation part 111.
[0148] Step 102 In complex arrangement generation part 112, About embedding series 913 generated by embedding series generation part 111, it is N. -- It divides into the real part and imaginary part of an element on one-dimensional complex arrangement, and is [ this ]. The and embedding complex pattern 921 are generated and it stores in storage part 150 of The 1. Complex arrangement One and The mention below in the details of generation part 112.
[0149] Digital watermark embedding device Embedding series generation part >
In embedding series generation part 111, it embeds by the following processings, and is Generate about series 913. To.
[0150] Generation of embedding series 913, For example, patent documents 1, "Takao Nakamura, Hiroshi Ogawa, Tomioka Jun*, Yoichi Takashima, Way method" 1999 year of the improvement in parallel translation' cutoff tolerance in digital watermarking A code and information security symposium, SCIS99-W3 - 2.1, pp.193 - 198 and 1999", "Takao Nakamura, Jun Katayama, Masashi Yamamuro, and the Sonehara Ascent -- "the analog picture or the high-speed digital-watermarking detection method of and others using a camera cell phone machine", being shown in IEICE TRANSACTIONS D-II, Vol. J87-D-II, No.12, pp. 2145-2155, and 2004" -- and To -- the constitution method of an embedding series [ like ] and the same method can be taken To. [0151] Below, it is as an example of embedding information 911,
* digital watermarking is embedded! /, To, t, and Example; in the case of obtaining and expressing only the fact
* Example in the case of being the information on lbit;
* The 1st example in the case of being the information on nbit;
* The 2nd example in the case of being the information on nbit;
it attaches [ it is alike and ] and explains. Other embedding which drops off in what is limited to these examples An intermediary is also good in the method of series generation.
[0152] (Example 1) As the numerical sequence to which embedding series 913 was expressed using the pseudorandom-numbers sequence, for example when embedding information 911 expressed only the fact "electronic Power is embedded" It may be calculated. Namely, pseudorandom-numbers sequence of average 0 (L is a system) rhonu= {rhonu, PN, - .., and PN}
1 W of 2 L Sequence}<img file="WO2007102403A1_D0023.tif" />
of -- it may determine like.
[0153] As a pseudorandom-numbers sequence, for example, M series and a GOLD series may be used.
[0154] (Example 2) When embedding information 911 is information on lbit, embedding series 913 is calculated as a numerical sequence which carried out spectral diffusion of the information on the lbit using the pseudorandom-numbers sequence, for example. It is good. That is, they are b and a pseudorandom-numbers sequence of average 0 about embedding information PN= {PN and PN --
1 2
PN the time of carrying out -- embedding series w= {w, w, --, w}} (L is the length of a series)
It carries out and is 1. 2 L
[0155] [32 number]<img file="WO2007102403A1_D0024.tif" />of -- it may determine like.
[0156] As a pseudorandom-numbers sequence, M series and a GOLD series may be used, for example.
[0157] (Example 3) When embedding information 911 is information which comprises nbit, it is embedding series 91. 3, For example, a pseudorandom-numbers sequence is used for every symbol which divided the information on the nbit for every mbit. It may be calculated as what multiplexed the numerical sequence which carried out The spectral diffusion. Namely, under Even if performed by procedure like an account, it is right .
[0158] a figure -- 14 is a flow chart of detailed operation of the embedding series generation part in the embodiment of The 1 of the present invention.
[0159] Divide embedding information 911 on step 201 nbit into a plurality of symbols S and S, - .., and S.
1 2 k Carry out. At this time, all the symbol may be [ every / the same mbit ], and each symbols differ. The information for the number of bits is expressed and and The are also right .
[0160] Here, a "symbol" expresses a part of information of the embedding information, respectively, the time of being the information used as the processing unit of actual electronic Penetration embedding, for example, being given with the binary number value of embedding information 911 power S64bit length -- a figure -- it is shown in 15 -- as -- every 12 bits information -- pause each of 12 bits of intermediaries' information is made to be made into a symbol -- ! and The -- good !. Certain !/, Is 1 symposium The information on lbit is denoted by Le and and The are also right . a figure -- embed like the example of 15 and the length (here 64 bits) of information 911 should divide among the length (12 bits) of each symbol! /and case, Part It may be made to carry out padding of a part of bit(s) of a symbol (here the last symbol) with a fixed value (here value 0).
[0161] Step 202 Spectral diffusion processing is performed to each symbol obtained at Step 201, and diffuse series P and - .. corresponding to each symbol, and P are generated.
1 k
[0162] The methods of spectral diffusion include the following methods, for example.
[0163] For example, when one symbol is expressing the information on lbit, they are each symbol S, ..., S.
1 K Receiving -- Respectively -- {1 and -- PN Series PN [ of K Kinds of Totals Which Take Value of 1} ] = (RhoEta, RhoEta, ...), and - '-Rho
1 11 12
k kl k2 when nu = (rhoeta and rhoeta, -- -) is generated and the value of a symbol expresses bit 1 to symbol i
the case where the value of a symbol expresses bit 0 for Is PN -- 1 -- PN is used as diffuse series P -- as -- even if -- it is good.
[0164] When one symbol is, for example, expressing the information which is 12 bits, it is a pair to each symbol. It carries out and prepares 4096 kinds of PN series PN, - .., PN, PN, - .., and PN, respectively.
(1,0) (1,4095) (2,0) (k,4095)
To symbol i, when the value of symbol i expresses integral value X with 12 bits, it is a diffusion system about PN. omega Even if it makes it use as sequence rho, it is right .
[0165] moreover -- as rhonu series (pseudorandom-numbers sequence) -- for example, -- even if it uses mu series and a GOLD series -- good,
[0166] Step 203 Diffuse series P power is also embedded and series w is calculated as follows. [0167], [a 33 number]
w -
For example, in the case of the latter example, it is as follows at the above-mentioned step 202.
[0168] [34 number]
(4 3)
<img file="WO2007102403A1_D0025.tif" />
In the above-mentioned formula,
[0169] [35 number]
I
**
the mosquito that [ whose ] of Is multiplied by is for making it the standard deviation of series each element set to 1, and the back if it embeds in calculation and intensity is controlled appropriately -- failing in Must -- also carrying out
[0170] [36 number]<img file="WO2007102403A1_D0026.tif" />
There is no Multiply by necessity.
[0171] (Example 4) When embedding information 911 is information which comprises nbit, it is embedding series 9. 13 uses pseudorandom numbers for one m times the length of this for the information on the nbit, and is direct spectrum Expansion, for example. It may be calculated as what was scattered. That is, it may be performed by the following procedures.
[0172] Embedding information 911 on 1 nbit was set to b, b .b, ..., b, and each bit was repeated m times respectively.
0 1 2 n
Series S is acquired. however, each bitb -- +1 and -- one value of one shall be taken
[0173] [37 number]
S = bob<sub>0</sub>---bob]b] "'b<sub>n</sub>- b<sub>n</sub>b<sub>n</sub>- -b*
m* m piece m*
2) S -- {-- +1 and -- being spread in pseudorandom-numbers sequence PN= {PN, PN, ..., PN} which takes 1} -- embedding
1 2 mn
Series w is searched for. namely, the time of considering it as w= {w, W, --, W},
1 2 mn
w =b PN b PN ---b PN (45)
1 0 1 0 2 0 m
(A lbit eye expresses b XPN and that 2nd bit is b XPN--.) It is the same as that of the following.
0 1 0 1
w =b PN b PN ---b PN w =bPN b PN ---b PN
mn n (m - l)n+l n (m - 1)n+2 n mn
The generation method of these embedding series [ like ] "Takao Nakamura, Jun Katayama, It is stated to Masashi Yamamuro and Sonehara Ascent "high-speed digital-watermarking detection [ of analog picture power using a camera cell phone machine ] method" IEICE TRANSACTIONS D- II, Vol. J87-D-II, No.12, pp.2145-2155, and 2004". To.
[0174] Digital watermark embedding device-complex arrangement generation part >
The processings following in complex arrangement generation part 112 of embedding complex pattern generation part 110a Complex pattern 921 is generated.
[0175] Operation of complex arrangement generation part 112 of complex pattern generation part 110a is shown below.
[0176] a figure -- 16 is a flow chart of operation of the complex arrangement generation part in the embodiment of The 1 of the present invention.
[0177] N of size M XM which is Value powerO of the element of all the step 301 -- One dimension
1 2 N-1
Complex number arrangement is prepared. However, M, M, ..., M take out two values respectively sequentially from [0178] step 302 embedding series 913 which is the number of elements defined beforehand, At Step 301 A value is set to arrangement so that the taken-out value may become a real part of the element value, and an imaginary part sequentially from the position of the prepared arrangement (0, 0, - .., 0). That is, complex arrangement is A [p, p, - .., and p] (p).
1 2 N-1 n
* When 0 and embedding series 913 are expressed as w, w, --, w,
1 Two It Carries Out.
A [0, 0, --, 0] =w+jw
1 2
A [l, 0, --, 0] =w+jw (46)
3 4
However, j is taken as an imaginary unit.
[0179] a figure -- this situation is shown in 17. however, a figure -- 17 The example in two-dimensional complex arrangement is shown.
[0180] Step 303 The complex arrangement generated at the above-mentioned step 302 is embedded, and it outputs as complex pattern 921.
[0181] It is N in this way. -- It is generating one-dimensional embedding complex pattern 921 as arrangement of the complex number based on embedding series 913 made using pseudorandom numbers, An embedding complex The value of each element of pattern 921 is determined are spread and distributed over complex number space. This It is spread so that the phase in embedding pattern 922 obtained as a result of time [ to mention below ] abnormal conditions may change with positions on N one-dimensional space, and bury at the time of detection of electronic Power. The size of the noise ingredient which originates in signal 912 before a lump and appears becomes smaller.
[0182] In advance of the procedure of the above-mentioned step 302, an order of embedding series 913 may be changed to a random order using pseudorandom numbers. The inaccurate analysis of the information currently embedded and an attack of rewriting etc. can be delivered difficult by this, and it is Inta. Local tolerance [ as opposed to / have an effect as Reeve coding and / embedding finishing signal 923 ] It contributes to preventing imbalance. In that case, kind of the pseudorandom numbers used for exchange of an order A value is given as a key of electronic Penetration embedding, and the same key is used at the time of detection of electronic Power. Carrying-out * This.
[0183] You may make it replace the element of the complex array obtained at Step 302 instead of replacing an order on embedding series 913.
[0184] Digital watermark embedding device 1-hour modulation part >
The details of operation of time modulation part 130 are explained below. [0185] Drawing 18 shows the example of composition of the time modulation part in the embodiment of The 1 of the present invention.
[0186] time modulation part 130a -- from periodic signal generating part 131, modulation part 132, and adder unit 133 -- composition -- Re and embedding complex pattern 921 are inputted and embedding pattern 922 is outputted.
[0187] the procedure of the following [ processing / embedding pattern 922 by time modulation part 130a / generation ] -- operation -- To be.
[0188] a figure -- 19 is a flow chart of operation of the time modulation part in the embodiment of The 1 of the present invention.
[0189] Step 401 In periodic signal generating part 131, two periodic signals which intersect perpendicularly with the same fundamental frequency are generated. For example, two periodic signals may be generated so that phases may differ by 90 degrees, i.e., 1Z4, cycle based on one periodic signal, respectively. Cycle to generate The example of a signal is mentioned below.
[0190] Step 402 It is in the direction of time, respectively by two periodic signals generated at Step 401 in the real part of inputted embedding complex pattern 921, and the imaginary part in modulation part 132. It modulates. The example of abnormal conditions is mentioned below.
[0191] Step 403 In adder unit 133, two N dimension signals modulated in modulation part 132 are added, embedding pattern 922 is acquired, and it stores in storage part 160 of The 2.
[0192] Explain the example of the periodic signal generated at Step 401.
[0193] The periodic signal generated by periodic signal generating part 131, Figure As shown in 4 A*C, it is the ((figure 4A) a) sine wave (figure 4B), the (b) triangular wave (figure 4C), and (c) rectangle wave, and it is as follows when a formula expresses each signal in the range of 0*t To T (T is a cycle of a signal).
[0194] [38 number]
(a) y = a ^ ox<img file="WO2007102403A1_D0027.tif" />It is as having already given details of these periodic signals.
[0195] Next, the abnormal conditions in modulation part 132 are N. -- It is a value of the real part of the value for every position of one-dimensional complex pattern 921, and an imaginary part, and is carried out by changing into the pattern of N dimension by and carrying out AM abnormal conditions by making into a subcarrier the periodic signal generated by periodic signal generating part 131.
[0196] That is, concrete for example, it is carried out as follows.
[0197] Now and N -- A one-dimensional complex pattern It shall be expressed with rho (chi, X, ..., X). This
1 2 N-1
The real part of P and an imaginary part shall be denoted by P and P the time of of.
(X, X, ..., X) rho (chi, X, ..., X) =P+j-P (x, chi, ..., chi)
1 2 N-1 R 1 2 Nu - 1I1 2 Nu - 1
(47)
It carries out. However, j is an imaginary unit.
[0198] Two periodic signal f^t f^t should be generated in periodic signal generating part 131.
. When f and f are here generated so that 90 degrees of phases may differ based on the same periodic signal It takes for an example,
f (t)=f (t-T/4) (48
It carries out. T is a cycle of a periodic signal.
[0199] It is f about rho and rho, respectively. (t) and f It modulates by (t) and obtains patterns W and W of N dimension with a following formula.
[0200] (chi, X, ..., X, t) W =P (X, X, ..., X) Xf (t)
r 1 2 N-1 r 1 2 N-1 r
(49)
W (x, x, ..., x, t) =P (x, x, ..., x) Xf (t)
i 1 2 N-1 i 1 2 N-1 i
(50)
In the above-mentioned step 403, mosquito Counting of this W and the W is carried out, and it is as embedding pattern 922. The following N dimension signal W is acquired.
[0201] W (xX, ..., X, t) =W (x, x, ..., chi, t)
1 2 N-1 r 1 2 N-1
+ W (x, x, ..., x, t) (51)
i 1 2 N-1
The example at the time of using a sine wave as a periodic signal is shown to the picture signal of N = 3 below.
[0202] f (t)=coscot (52)
f (t) =sin ot If (53),
(x, y) W(x, y, t) =P cos ot+P (x, y) sin omega t (54)
It becomes. However, omega is omega=2pizetatau in the angular velocity corresponding to cycle T.
[0203] It is a function of complex variable as a function by which two periodic signals are denoted for calculation of these.
i)<sup>=</sup>e =cos ot+jsin ot
Using -- until style trap and a thing call it a line as follows -- ,
[0204] [39 number]
W(x, y, *) = niPix and yx/(*) (t) ] = ^rho(chi, y) e'<sup>eat</sup>] (55)
However, f<sup>W</sup>(t) is f. Conjugate complex number of (t),
[0205] [40 number]
It is an operation which takes out the real part of Is C. It comes out.
[0206] <electronic watermark embedding device Embedding pattern superposed part >
Below, the details of operation of embedding pattern superposed part 140 are explained.
[0207] With embedding pattern superposed part 140, It is generated by time modulation part 130a to N dimension signal inputted as signal 912 before embedding, Nth order stored in storage part 160 of The 2 N dimension signal of the result which was superimposed by adding the original embedding pattern 922, and was superimposed It outputs as embedding finishing signal 923.
[0208] When adding embedding intensity embedding pattern 922, he emphasizes with predetermined intensity parameter a, and may be trying to be embedded. Namely, Trust before embedding of N dimension Item 912 It is W (x, chi, ..., chi, t) about iota (chi, X, ..., X, t) and embedding No turn 922.
1 2 N-l 1 When N-l Two, it is about Embedding Finishing Signal 923 (Chi, Chi, ..., Chi, T),
1 2 N-l
(chi, X, --, X, t) = I(x, X, --, X, t)+a -W (x, x, - - -, x, t) (56)
1 2 N-l
of -- it asks like.
[0209] Intensity parameter alpha may be constituted so that it may change according to the before [ embedding ] signal 912 whole, and the characteristic quantity which are made into the object of an operation among signals 912 before embedding and which is part Power ゝ computed. for example, the case where signal 912 before embedding is a picture signal -- frame picture the embedding pattern, in which the texture field, the intense field of the motion, etc. were added -- being conspicuous -- a stake part -- strong That is, it is common [ of a frame picture ] so that alpha may take a big value. About conspicuous parts, such as a Rugged field and a field of a slow unific motion It is constituted so that embedding may be performed weakly (that is, a takes a small value like). It does not matter.
[0210] In the case of superposition of the repetition embedding pattern of an embedding pattern, the size of signal 912 before embedding embeds, size comes rather than the size of pattern 922, and and a case may be added so that embedding pattern 922 may be repeated.
[0211] The size of signal 912 before embedding embeds and there is the following, for example as an example in the case of being larger than the size of pattern 922.
[0212] In the case where signal 912 before 1 embedding is a picture signal, signal 912 before embedding The length (frame number) of the direction of time embeds, and longer than the length of the direction of time of pattern 922 a case -- Drawing 20 -- like -- embedding pattern 922 -- two or more times -- repeating -- it may make .
[0213] In the case where signal 912 before 2 embedding is a picture signal, Signal 912 before embedding The size (field angle) of a frame picture embeds and it is a larger place than the size (field angle) of pattern 922. As shown in Synthesis and Drawing 21, it adds so that it may cover with embedding pattern 922 in the shape of a tile in all directions. It may carry out.
[0214] It may be made to expand embedding pattern 922 in advance of expansion of an embedding pattern, and superposition of an embedding pattern, so that it may be equal to arbitrary sizes or the size of signal 912 before embedding. Figure An example is shown in 22. Figure It embeds with the example expanded twice in 22, and is before Trust. It is good even if it is expanded to other sizes, although the example expanded according to the size of item 912 was shown. It is and things cannot be overemphasized.
[0215] What kind of algorithm may be used for expansion. a figure -- even if it is made for one value to correspond to expanded 1 block like 22 and is made to make all blocks after expansion into the same value -- right -- -- carrying out -- an alignment complement and by -- it may be made to use any cubic publicly known interpolation technique
[0216] a figure -- until it is made as [ expand / the example which expands a frame picture to a unit in the direction of space is shown, and / when signal 912 before embedding is a picture signal / in the direction of time / in 22, / and To ] and style trap and a thing also say and The -- ,
[0217] When it embeds to the embedding to characteristic quantity, and signal 912 before embedding and pattern 922 is superimposed, The value of direct embedding pattern 922 is added to the signal value of signal 912 before embedding. Instead of carrying out, It embeds so that the predetermined characteristic quantity of signal 912 before embedding may embed and only the value of pattern 922 or its multiplication by scalars may be changed, and last signal 912 is changed. Intermediary superposition may be performed.
[0218] As an example of the above-mentioned characteristic quantity, there are a signal value of signal 912 before embedding, average value of the signal value for every block, etc., for example. signal 912 before embedding -- case of a picture signal or an image signal To -- even if it, for example, uses the luminosity value of the pixel of an image and a picture, color difference, the color signal value of RGB, etc. It is good.
[0219] Digital watermark detecting device >
It comprises time demodulation section 210, detection information extraction part 220, and No turn storage part 250, embedding finishing signal 923 is inputted, and digital-watermarking detecting device 200 shown in Drawing 10 is detection information 91. 4 is outputted.
[0220] Below, explain operation of digital-watermarking detecting device 200.
[0221] a figure -- 23 is a flow chart of operation of the digital-watermarking detecting device in the embodiment of The 1 of the present invention.
[0222] Step 501 In time demodulation section 210, based on inputted embedding finishing signal 923, it restores to the direction of a time-axis, detection complex pattern 961 is acquired, and it stores in pattern memory part 250. N which, as for detection complex pattern 961, complex number power also comprises -- One-dimensional putter It is A. The details of operation of time demodulation section 210 are mentioned below.
[0223] In addition in advance of the time recovery processing by time demodulation section 210, for example, geometric modification amendment, noise rejection, filtering, block superposition, blocking, etc. may be processed by pretreatment to embedding finishing signal 923. These examples are mentioned below.
[0224] Step 502 In detection information extraction part 220, it being obtained by time demodulation section 210, analyzing detection complex pattern 961 stored in pattern memory part 250, carrying out electronic Penetration, and burying -- Included The electronic Penetration information which saw and was embedded with device 100 is extracted, and it outputs as detection information 914.
[0225] Mention the details of operation of detection information extraction part 220 below.
[0226] In addition, precede the detection information extraction processing by detection information extraction part 220, and it is N by pretreatment to detection complex pattern 961. -- Geometric modification amendment in one-dimensional space, noise rejection, filtering, block superposition, blocking, etc. may be processed. It is One to these examples. If it is, it mentions below.
[0227] Pretreatment > to a Embed finishing signal
Front place to embedding finishing signal [ in / to the following / digital-watermarking detecting device 200 ] 923 The example of Reason is shown.
[0228] As opposed to embedding finishing signal 923 with which digital watermarking was embedded in geometric modification amendment electronic watermark embedding device 100, When geometric modification of expansion, reduction, rotation, parallel translation, aspect ratio change, projection conversion, etc. is added, detection of electronic Power is difficult as it is. It is right even if it performs the pretreatment which amends this, since it may become.
[0229] Geometric strange * amendments are literature "Csurka, G., Deguillaume, F., Ruanaidh, and J.J, for example. K. O , and Pun, T., A Bayesian Approach to Affine Transformation Resistant Image and Video Watermarking," Information Hiding, It is shown in Proceedings Lecture Notes in Computer Science 1768, pp.270-285, and Springer- Verlag (2000)", Electronic penetration Aside from the information on Power, the signal for performing geometry amendment is embedded to the signal, and it is This. The grade of the change added to the signal by detecting Re is presumed, and it is presumed to a signal. It may amend by performing inverse transform of Modification. For example, Literature "Honsinger, C, "Data embedding using Phase Dispersion, iota*epsilon Seminar on secure Images and Image Authentication (Ref.No.) 2000/039, and pp.5/1- it is shown in five sevenths (2000") -- as -- electron The embedding pattern itself which has the information which carries out With power and it embeds Circumference with a repetition A term pattern is used and it is autocorrelation-function power at the time of detection. Change of the cycle is observed. It is based on expansion and reduction percentage obtained by computing expansion and reduction percentage, and is Application about the inverse transform of change. It may amend by Surely. the target signal for example, -- the case of a picture or an image -- sentence Dedication -- "-- i application high-speed corner detection algorithm "and IEICE TRANSACTIONS D=II for Jun Katayama, Takao Nakamura, Masashi Yamamuro, Noboru Sonehara, and digital-watermarking reading, and Vol. J88-D-II, No.6, and pp.1035- 1046 and 200 extracting the square area in a picture by a method as shown in 5" -- there -- Electronic transparent ゝTo -- Fill -- Included -- rare -- I will -- As -- geometric amendment is made to be performed -- Please -- good !.
[0230] As opposed to embedding finishing signal 923 with which filter processing electronic Power was embedded, The filter processing which a noise is added and removes a noise to To go case, and signal ingredient of digital watermarking Mosquito Scare is also good in the filter processing which can remove the ingredient of the original signal before embedding, leaving.
[0231] In block superposition electronic watermark embedding device 100, The size of signal 912 before embedding embeds and size comes rather than the size of pattern 922. It is added to a case so that and Embed pattern 922 may be repeated to embedding pattern superposed part 140, and it is a and To case, burying -- Included carving for every part of the pattern which it saw [ pattern ] and had settled signal 923 repeated -- it is superposition to one about them Block superposition processing which is added and is summarized to one block may be performed.
[0232] In blocking electronic watermark embedding device 100, The size of signal 912 before embedding embeds, and when larger than the size of pattern 922, it is embedding pattern superposed part 1. In 40, When embedding pattern 922 is expanded, it embeds. After reducing finishing signal 923, electronic Power is made to be detected, and Go is also good !.
[0233] In detection electronic Penetration embedding device 100 from characteristic quantity, When it embeds to signal 912 before embedding and pattern 922 is superimposed, The predetermined characteristic quantity of signal 912 before embedding Value of embedding pattern 922, or -- embedding so that only multiplication by scalars may be changed -- before Trust Finishing of embedding when it is made to be superimposed by changing item 912 based on what computed predetermined characteristic quantity, electronic Power is detected from signal 923 -- as -- even if -- it is good.
[0234] for example, a figure -- 24 -- like -- embedding finishing signal 923 -- carving 1 for every block -- each Blog The characteristic quantity of A is computed, the sequence of characteristic quantity is constituted, and it may be made to detect electronic PenetrationゝTo from now on. As an example of characteristic quantity, the average value of the signal value in a block may be used, for example. In the case of a picture signal or an image signal, signals are an image and a picture, for example. Even if it uses the luminosity value of a pixel, color difference, the color signal value of RGB, etc., it is right .
[0235] About the pretreatment mentioned as these examples, it is a front place of the detection information extraction processing by detection information extraction part 220. As Reason, it may be made to carry out to detection complex pattern 961. especially -- geometric modification amendment about -- case where the axis of eye N dimension is the direction of time in a picture signal, Finishing of embedding Elasticity of the direction of time is amended as a pretreatment to signal 923, and it is to detection complex pattern 961. By performing geometric modification amendment of the direction of space as a receiving pretreatment, it is empty with efficiently and sufficient accuracy. It is possible to perform geometric modification amendment [-like / INL / in between ].
[0236] Digital watermark detecting device 1-hour demodulation section >
Below, the details of operation of time demodulation section 210 are explained.
[0237] a figure -- 25 shows the example of composition of the time demodulation section in the embodiment of The 1 of the present invention.
[0238] Time [ to be shown in the figure ] demodulation section 210 comprises periodic signal generating part 211, demodulation section 212, and complex pattern formation part 213, inputs embedding finishing signal 923, and outputs detection complex pattern 961.
[0239] in addition -- a figure -- in 25 -- a figure -- which composition was indicated that information flows upwards from the bottom, and was noticed about and To in order to make an understanding of correspondence with 18 easy,
[0240] the procedure of the following [ processing / of embedding finishing signal 923 by time demodulation section 210 / recovery ] -- operation -- To be.
[0241] a figure -- 26 is a flow chart of operation of the time demodulation section in the embodiment of The 1 of the present invention.
[0242] Step 601 In periodic signal generating part 211, two periodic signals which intersect perpendicularly with the same fundamental frequency are generated. For example, two periodic signals may be generated so that phases may differ by 90 degrees, i.e., 1Z4, cycle based on one periodic signal, respectively. Circumference to generate Time modulation part 130a in the above-mentioned electronic watermark embedding device 100 is provided with a term signal. It corresponds to periodic signal generating part 131. It is to already have described the example of the periodic signal. It is a cage.
[0243] Step 602 It gets over by each of two periodic signals generated at Step 601 in demodulation section 212 based on the direction ingredient of time of inputted embedding finishing signal 923, and is two. N -- A one-dimensional signal is acquired. The example of a recovery is mentioned below.
[0244] Step [ Detection complex pattern 961 which is a pattern is obtained. ] 603 Two N to which it restored in demodulation section 212 in complex pattern formation part 213 -- N of the complex number from which it was made for a one-dimensional signal to serve as a real part and an imaginary part, respectively -- One dimension
[0245] Being concrete and two N -- They are Q (--- x, x, x) and Q (x, x, *) about a one-dimensional signal, respectively.
r 1 2 N-l i 1 When referred to as 2 x, it is considering detection complex pattern 961 as Q (x, chi, - chi),
N-l 1 2 N-l
Q(x, x, "X) =Q (x , x , "X)
1 2 N-l r 1 2 N-- 1
+jQ (x , x , ---x) (57)
i 1 2 N-l
It comes out and asks. However, j is an imaginary unit.
[0246] Here explains the example of a time recovery.
[0247] Periodic signal generation of embedding finishing signal 923 of the recovery in demodulation section 212, and N dimension It is carried out by asking for the phase of the periodic signal generated in part 211. It is 2 especially as follows. By asking for the size of those ingredients from the periodic signal of One, it is Capacity about the phase of a periodic signal. It can measure to Easy.
[0248] That is, it is specifically, for example, as follows, carried out.
hereinafter, embedding finishing signal 923 of N dimension it is expressed with gamma (chi, X, ..., X, t) -- Listen
1 2 N-1
It is considered as of.
[0249] In periodic signal generating part 211, they are two periodic signals f. (t) and f That by which (t) was generated It carries out. Here, ^ was generated so that 90 degrees of phases might differ based on the same periodic signal. A case is taken for an example,
f (t) =f (t-T/4) (58)
It carries out.
[0250] It is f about gamma. (t) and f It gets over by (t), and is a following formula, and they are two N. -- One-dimensional signals Q (X, X, ..., X) and Q
r i r 1 2 N_l i
X, X, ..., X, and t are obtained.
1 2 N-l
[0251] [41 number] Q^ -- '- =/' (xi, x)<sub>2</sub>'--' ^^ 'r (59) rho and omicron 1 =<sup>2</sup>' -- '1'/'<sup>6</sup>0)
Here, the section which is embedded and is made into the object of detection among ending signals 923 each starts t and t.
1 2
They are a point and an ending point. For example, it is a detection pair in all the inputted embedding finishing signals 923. It is t = so that it may be considered as an elephant. -- Embedding finishing signal 923 good also as infinity and t =infinity and inputted
1 2
a part for a of n cycle is taken out -- as -- t = 0 and t =nT (however, T -- periodic signal generating part 211 -- generation)
1 2
It is good also as a cycle of a To be cycle signal.
[0252] Embedding finishing signal 923 is acquired as a discrete signal, and it is two by the following Sum of products calculations in a and To case. nu -- One-dimensional signals Q and Q may be searched for.
[0253] [42 number]
2 ,*<sub>2</sub>,..'';^-1)=2^'(2 ''^-)) (<sup>6</sup>D
r=ti
^ (A , --, ^ , A, --) /! " (62)
tau=
For example, in a picture signal etc., processors of low performance, such as a mobile phone, are used. When the re-photoed image power also detects electronic Power, the frame rate of photography is stable. The timing of a sampling may shift minutely, without carrying out. In such a case, signals gamma (chi, X, ..., X, 1) and gamma (chi, X, ..., X, 2) and ... which were obtained by discrete , gamma (chi and chi)
1 2 Nu-1 1 2 Nu-1 1 2
using detection time t and t of each signal, ..., t to ..., chi, and eta -- as follows
N-1 1 2 n
It is origin to the i-th sample about a product with value [ of the periodic function in the i-th measurement time ] f (t). By doing Sum of products calculation, an unstable frame rate is amended and the accuracy of a calculation result is maintained. It can do. [0254] [43 number]
<img file="WO2007102403A1_D0028.tif" />The images displayed on a screen, TV, etc. are cameras, such as a video camera and a mobile phone. When a photograph is taken, the information about a reproductive frame rate and the frame rate of photography obtains , -- Being done -- a case -- signal which was most, and was discretely acquired when asking for t using it * Use gamma (X, eta).<img file="WO2007102403A1_D0029.tif" />
It can perform The and calculating as follows.
[0255] [44 number]
(6 5)
, F<img file="WO2007102403A1_D0030.tif" />
However, F is a frame rate at the time of photography.
[0256] Next, explain the time recovery using difference' differentiation. Instead of the above, the difference or differentiation of the direction of t axis of embedding finishing signal 923 gamma (chi, X, - - -, X, t) is used, and it is Demodulate.
1 2 N- 1
It is good as for a method of To. the example of composition of such a time demodulation section 310 -- a figure -- it is shown in 27.
[0257] a figure -- time [ to be shown in 27 ] demodulation section 210b -- a figure -- time demodulation section 210a of 25 -- abbreviated -- after Am and embedding finishing signal 923 with the same composition were inputted into - A signal differentiation part 215 -- demodulation section 2
It is constituted so that it may be inputted into 12, and differs from Point.
[0258] as opposed to embedding finishing signal 923 inputted in signal differentiation part 215 -- calculating the difference or differentiation of the axis of t axis of eye N dimension, i.e., the direction, and making it outputted to demodulation section 212 It is.
[0259] For example, when a picture signal etc. are considered, it is known that inter-frame correlation of a signal is large. If inter-frame correlation calculates inter-frame difference and a differential value since it is high so that the time interval of two frames is short, more than the offset part by a periodic signal -- original picture ingredient it will be canceled greatly and the energy of electronic Power will increase relatively -- this -- detection of digital watermarking becomes easy by recovering from a difference value [ like ] and a differential value -- detecting accuracy There is an advantage of improving. On the contrary, it is a said degree even if it is the embedding of weaker digital watermarking. Since the detection performance of a degree can be maintained, less electronic Power of degradation of a signal buries. A lump is possible.
[0260] Digital watermark detecting device Detection information extraction part >
Next, detection information extraction part 220 is explained in detail.
[0261] a figure -- 28 shows the example of composition of the detection information extraction part in the embodiment of The 1 of the present invention.
[0262] It comprises detection series extraction part 221, correlation value calculating part 222, maximum judgment part 223, and detection information reconstruction section 224, and detection information extraction part 220a is detection complex The from pattern memory part 250. Turn 961 is inputted and detection information 914 is outputted.
[0263] in addition -- a figure -- in 28 -- a figure -- which composition was indicated that information flows upwards from the bottom, and was noticed about and To in order to make an understanding of correspondence with 12 easy,
[0264] Detection information extraction processing by detection information extraction part 220a is carried out in the following procedures.
[0265] a figure -- 29 is a flow chart of operation of the detection information extraction part in the embodiment of The 1 of the present invention.
[0266] Step 701 It is a style about detection series 1113 which took out and put the value of the real part and the imaginary part in order in detection series extraction part 221 from the inputted detection complex pattern 961 power profitable To be complex number value. Formation is carried out. The details of operation of detection series extraction part 221 are mentioned below.
[0267] Step 702 Phase with the embedding series constituted based on the embedding series assumed to be detection series 1113 which comprised detection series extraction part 221 in correlation value calculating part 222 Seki is calculated and correlation value 1114 is calculated.
[0268] When the value which changes with kinds of embedding series is embedded, shave correlation with a plurality of embedding series constituted based on a plurality of embedding series considered. Calculation is carried out and corresponding correlation value 1114 is calculated. the details of operation of correlation value calculating part 222 -- just -- The mentions below.
[0269] Step 703 Correlation value acquired by correlation value calculating part 222 in maximum judgment part 223
The correlation value calculating part corresponding to correlation value 1114 which 1114 finds the thing used as the maximum and serves as the maximum
The embedding series used by the correlation computation of 222 is determined.
[0270] A constitution method of an embedding series [ in / in addition / electronic Penetration embedding device 100 ]
It is made to be judged by other methods instead of the maximum judging by maximum judgment part 223. Me me is also good.
[0271] the details of operation of maximum judgment part 223, and the details of other methods used as substitution -- back -- Description is carried out.
[0272] Step 704 It is a re-style about detection information 914 it is judged that it was actually embedded in detection information reconstruction section 224 based on the embedding series determined by maximum judgment part 223. Formation is carried out. The details of operation of detection information reconstruction section 224 are mentioned below.
[0273] Digital watermark detecting device Detection information extraction part Detection series extraction part >
Next, the details of the above-mentioned detection series extraction part 221 are explained.
[0274] Detection series extraction part 221 manages the function by the side of the detection corresponding to complex arrangement generation part 112 in complex pattern generation part 11Oa of electronic Penetration embedding device 100, and is a detection complex putter. Complex number value power acquired from A 961 Detection series 1113 is constituted.
[0275] Processing by detection series extraction part 221 is carried out in the following procedures.
[0276] a figure -- 30 is a flow chart of detailed operation of the detection series extraction part in the embodiment of The 1 of the present invention.
[0277] Step 801 From detection complex pattern 961 to a size M X M chi - .. chi mu nu -- 1st order
1 2 N - One The original complex arrangement is constituted and it stores in a memory (not shown). However, mu, mu, ..., mu
1 2 N - One Electronic Penetration It is the number of elements same with having used by complex arrangement generation part 112 of embedding device 100.
[0278] When detection complex pattern 961 is obtained as a discrete signal, it is it as it is N--1 It is considered that it is the complex arrangement of a dimension. Detection complex pattern 961 is obtained as a continuation signal. It is N about what carried out specimen Y of the detection complex pattern 961 using arbitrary specimen Means when it was. -- It uses as one-dimensional complex arrangement.
[0279] complex arrangement power acquired from the step 802 memory (not shown) at Step 801 the real part of the complex number value which took out every one complex number value in order, and took it out, and an imaginary part -- respectively -- independent -- real -- It sees as a numerical value and arranges. That is, complex arrangement is A [p, p, ..., p] (p*0). Detection series 1
1 2 N- 1 n
When expressed 114 power and ...,
1 2 L
[0280] [45 number]
<img file="WO2007102403A1_D0031.tif" />(6 8)
14 = 3 [1, 0, -
however
[0281] [46 number]
iR, 3
It is an operation of Is and a complex number which takes out a real part and an imaginary part, respectively. It comes out.
[0282] This is generation of complex arrangement generation part 112 of electronic Penetration embedding device 100, and symmetrical processing.
[0283] Step 803 Obtained i , gamma, ..., gamma are outputted as detection series 1113. moreover
1 Two It Carries Out.
In complex arrangement generation part 112 of electronic Penetration embedding device 100, composition of complex arrangement preceding and using pseudorandom numbers for an order of embedding series 913 -- a random order -- exchanges Case where Have been embedding is made The step 803 concerned is preceded and it is detection series 1113. It returns to an order which , embedding series 913, and correspondence attach by changing an order to the time of complex arrangement generation part 112, and a contrary using pseudorandom numbers. In that case, it is To use to exchange of an order. The value of the kind of Pseudo random numbers is business as a key to digital-watermarking detection at the time of the embedding of digital watermarking. The same key as the ability to have been is given. [0284] In complex arrangement generation part 112 of electronic watermark embedding device 100, Embedding system The element of the obtained complex array is replaced instead of replacing an order on sequence 913. When carrying out by To, Complex distribution constituted from Step 801 in advance of Step 802 It may be made to return the element of a sequence by changing to the time of complex arrangement generation part 112, and a contrary.
[0285] Digital watermark detecting device Detection information extraction part Correlation value calculating part >
Next, it is To explain about the details of operation of correlation value calculating part 222 of detection information extraction part 220a. To.
[0286] Correlation value calculation processing in correlation value calculating part 222 is carried out in the following procedures.
[0287] One Correlation value calculating part 222 is an embedding series generation part of electronic watermark embedding device 100.
Embedding series w considered by the same procedure as 111<sup>(1)</sup>, w<sup>(2)</sup>-- is generated.
[0288] the embedding series considered being all the embedding series which may be embedded here, and responding to generation of the embedding series in embedding series generation part 111 -- for example, -- It can decide as follows.
[0289] (Example 1) For example, the embedding series considered when it embeds like (Example 1) of embedding series generation part 111 and the series is generated,
w=PN= {PN, PN, - .., and PN} (69)
1 2 L
It is one kind.
[0290] (Example 2) For example, the embedding series considered when it embeds like (Example 2) of embedding series generation part 111 and the series is generated,
w<sup>(1)</sup>= PN= {PN, PN, - .., and PN} (70)
1 2 L
w<sup>(2)</sup>= -PN= {-PN, -PN, - .., and -PN} (71)
1 2 L
They are two kinds of of.
[0291] (Example 3) For example, each symbol expresses the information on lbit in (Example 3) of embedding series generation part 111! /and the embedding series which it embeds like the example of To, and a series is generated, and is considered at a and To case should receive each symbol i.
w<sup>(i</sup>'= rhonu = {rhoeta, rhoeta, - .., and rhoeta} (72)
i il i2 iL
w<sup>(i</sup>'<sup>2)</sup>= - PNi= {-pn, -pn, - .., and -- pn} They are two kinds of (73).
[0292] Each symbol expresses the information which is 12 bits in (Example 3) of embedding series generation part 111. When it embeds like the example of ing and the series is generated, the embedding series considered receives each symbol i,
w<sup>(i</sup>'= rhonu (74)
(i, 0)
(i, 2) _<sub>pialphatau</sub>
w -- PN
(i, 1)
(i, 4096) _
w -- PN
(i, 4095)
They are 4096 kinds of of.
[0293] (Example 4) For example, it embeds like the example of (Example 4) of embedding series generation part 111, and a series is generated! The embedding series considered to cover all the cases simply in a /To go case is 2.<sup>n</sup>It becomes a passage. In next correlation computation, it correlates by detection series 1113 which divided m detection series 1113 at a time, and divided correlation with two kinds of embedding series as follows to one bit b. It may be made to calculate.
[0294] [47 number]
w<sup>(</sup>',') = {+ PN<sub>m+]</sub>,+PN<sub>im+2</sub>'--} (7 5) w ('.)<sup>2</sup>) = {-- PN<sub>m+]</sub>,+PN<sub>im+2</sub>-- and - Up } (7 6)
High-speed digital-watermarking detection method "of analog picture power in which above-mentioned literature "Takao Nakamura, Jun Katayama, Masashi Yamamuro, and cellular phone with a Sonehara Ascent" camera were used for such a method It is indicated to IEICE TRANSACTIONS D-II, V ol. J87-D-II, No.12, pp. 2145-2155, and 2004."
[0295] Two Detection series 1113 acquired by detection series extraction part 221, and each embedding series w acquired by above 1<sup>(1)</sup>, w<sup>(2)</sup>-- Correlation is calculated, respectively.
[0296] Correlation computation is [ patent documents 1 and ] literature "Takao Nakamura, for example, Way method "of the improvement in parallel translation' cutoff tolerance which can be put on Hiroshi Ogawa, Tomioka Jun*, and Yoichi Takashima" digital watermarking The 1999 code and information As [ show / in security symposium and SCIS99-W3-2.1, pp.193-198, and 1999" ] To the same for example, Sum of products operation like a following formula as what is performed by detection of electronic Be sought after. It is detection series 1 about a correlation value to ask for this, and gamma= {gamma, gamma, ..., gamma}.
1 Two It Carries Out.
113} is made into the target embedding series now.
<img file="WO2007102403A1_D0032.tif" />
[0297] [48 number]<sub>=i</sub>-.<sub>w</sub>)<sub>=</sub>V,) (7 7)
Jt=l
However, '" is an inner product operation when a sequence of numbers is regarded as a vector.
[0298] above-mentioned literature "Takao Nakamura, Jun Katayama, Masashi Yamamuro, and the Sonehara Ascent -- "high-speed digital-watermarking detection method of analog picture power using a camera cell phone machine", In order to arrange the valuation basis of IEICE TRANSACTIONS D-II, Vol. J87-D -II, No.12, pp.2145-2155, and detection reliability that is in 2004", gamma For example, reach. The following [ it carries out regular Y of each element so that it may become average 0 and distribution 1 beforehand ] Even if it calculates by multiplying by an absolute term by correlation value calculation like, it is style trap .
[0299] [49 number]
p persons -- = 1 (78),
Digital watermark detecting device Detection information extraction part Maximum judgment part >
Next, maximum judgment part 223 of detection information extraction part 220a is explained in detail.
[0300] Processing in maximum judgment part 223 is carried out in the following procedures.
[0301] Correlation value 1114, p which were obtained by one correlation value calculating part 222<sup>(2)</sup>A value serves as the maximum from --. Correlation value It is found.
[0302] [50 number]<img file="WO2007102403A1_D0033.tif" />
However, it is an operation which returns MAX0 and Is the maximum value.
[0303] Two Embedding series w corresponding to a plate<sup>max</sup>It obtains.
[0304] The greatest correlation value p<sup>(max)</sup>judging whether it is over the threshold of Is predetermined -- predetermined Threshold exceeding [ ] a value -- electronic Power is embedded at and a case, and it is judged as and a It is a force thing -- as -- even if -- it is good.
[0305] Below, explain operation used as substitution of maximum judgment part 223.
[0306] Embedding series w of With which is correlation value calculating part 222 instead of the maximum judging by maximum judgment part 223<sup>(1)</sup>, w<sup>(2)</sup>It is embedding series w, without calculating correlation to --.<sup>(1)</sup>It correlates with In order. The correlation value acquired by calculating judges whether it is over the predetermined threshold, and exceeds a threshold. It is w about a Embedding series.<sup>(max)</sup>It is good, even if it carries out and correlation computation is ended at the time. It is and is ..
[0307] Embedding series power in electronic watermark embedding device 100 For example, as [ showed / (Example 1) of embedding series generation part 111 ] -- comprising only one kind of embedding series -- Fill -- Included -- rare -- ing -- a case -- To -- since only one is calculated as for a correlation value, there is no meaning in the maximum judging by maximum judgment part 223. It is made to be judged by whether they are V and To exceeding a predetermined threshold, and and The are [ the correlation value acquired instead ] also right .
[0308] Embedding series power in electronic watermark embedding device 100 For example, as [ showed / (Example 2) of embedding series generation part 111 ] -- the difference in the positive/negative of one kind of embedding series -- style Case where it is accomplished and embedded Since the correlation value which numerals reversed is calculated, it is a phase. The correlation computation in the Seki value calculation part 222 is a deed and the maximum only about one embedding series. Instead of the maximum judging by judgment part 223, it is embedded with the numerals of the acquired correlation value. A It was value is judged. Threshold predetermined in the absolute value of the acquired correlation value It is made to be judged by whether they are and To exceeding a value, and ! and The are also good !.
[0309] It spaces in the size of a correlation value and may make it evaluate the reliability of detection.
[0310] Digital watermark detecting device Detection information extraction part Detection information reconstruction section >, next detection information reconstruction section 224 of detection information extraction part 220a are explained in detail.
[0311] Processing in detection information reconstruction section 224 is carried out in the following procedures.
[0312] One Embedding series w acquired by maximum judgment part 223<sup>(max)</sup>It carries out in Or et al. and Electronic penetration, and buries. It responds to the embedding series generation method in embedding series generation part 111 of lump device 100, and is a pair. The value of the embedding information which carries out Response is constituted as detection information. For example, embedding series 913 Comprising a form which carries out spectral diffusion, !To go case is w.<sup>max</sup>It is Inspection in the form which carries out reverse spectral diffusion. Appearance information is reconstructed.
[0313] (Example 1) For example, like (Example 1) of embedding series generation part 111, when the embedding series is generated, the information itself called Powerlessness where electronic Power was embedded turns into detection information.
[0314] (Example 2) Like (Example 2) of embedding series generation part 111, when the embedding series is generated, it is w.<sup>(max)</sup>= If it is PN, detection information is bit value 1 w.<sup>(max)</sup>= -- It is that at PN. Detect information serves as bit value 0.
[0315] (Example 3) For example, each symbol expresses the information on lbit in (Example 3) of embedding series generation part 111! It embeds like the example of /and To, a series is generated, and a and To case is received at symbol i,<sub>w</sub><sup>(max)</sup>= A symbol value is 1 w if it is PN.<sup>(max)</sup>= -- from which a symbol value will be set to 0 if it is PN. It is Detection information by connecting the symbol value acquired by performing this to all the symbols. News is obtained.
[0316] It is [ as opposed to / when each symbol embeds like the example expressing the information which is 12 bits and the series is generated in (Example 3) of embedding series generation part 111 / symbol i ] w.<sup>(max)</sup>= A symbol value will be set to x if it is PN. It is a line to all the symbols about this! It is obtained.
(i,x)
Detection information is acquired by connecting a It was symbol value.
[0317] (Example 4) For example, like the example of (Example 4) of embedding series generation part 111, when the embedding series is generated, receive bit b.
w) = {+rhonu, +PN, - .., and +PN}
im+1 im+2 im+m
If come out and it is, it will be bit value +. 1
w()= {-- PN, -PN, - .., and -PN}
im+1 im+2 im+m
It will be a bit value, if come out and it is. -- It is set to 1. Detection information 914 is acquired by connecting the bit value acquired by performing this to all the bits.
[0318] Two Detection information 914 acquired above is outputted.
[0319] <How to embed others to a complex number>
next, how [ to embed others to a complex number ] Stickiness [ ] -- it Explanation.
[0320] In the above-mentioned example of composition, electronic Penetration embedding device 100 smells complex arrangement generation part 112. When setting up the element value of complex arrangement from The and embedding series 913, it is from embedding series 913. It was made for the taken-out value to become a real part of an element value, and an imaginary part. It corresponds to this and is Electronic penetration. It lends and detects from detection complex pattern 961 in detection series extraction part 221 of detecting device 200. When series 1113 is constituted, Value of the real part of the complex number value of detection complex pattern 961, and an imaginary part Detection series 1113 consisted of taking out and arranging.
[0321] Although the real part and imaginary part of the complex number were used in the example of these, it is complex arrangement generation part 11. Even if it will use a complex number value so that these may differ if operation of 2 and detection series extraction part 221 is matched, it is style trap .
[0322] For example, values w and w taken out from embedding series 913 in complex arrangement generation part 112
1
This may be set up to become the angle of deviation and the absolute value of the element value of complex arrangement when referred to as 2. In this case, in detection series extraction part 221, it is side of the complex number value of detection complex pattern 961. It will be right if detection series 1113 is constituted using an angle and an absolute value.
[0323] For example, one value w taken out from embedding series 913 may be made to correspond to the angle of deviation of one complex number value, and may also be embedded. At this time, the absolute value of a value is to 1 two or more [ more than ], for example. It may fix. for example, the value of an embedding series -- + 1 and 1 -- the time of taking either of 1 -- complex the numerical angle of deviation -- respectively -- It may be made to be referred to as pi zeta4 and 3 pi zeta4. In this case, detection series and The use only the angle of deviation of the complex number value of detection complex pattern 961 for extraction part 221, and it is Inspection. Appearance series 1113 may be constituted. such a constitution method -- a Ivy case -- complex arrangement -- complex constituting as arrangement of the real value showing the value of the angle of deviation instead of calculating as arrangement of a number -- this -- even if it constitutes time modulation part 130 so that the phase of a periodic variable may be controlled using a value -- right , This When a value is set to the real part and imaginary part of a complex number in a constitution method [ like ] as for a Ivy case Comparing, the length of an embedding series becomes half.
[0324] It is [ as opposed to / for example, / w and w ] an element value of complex arrangement by the following formulas. rho is determined. It is good.
[0325] p=aw +bw (79
1 2
It is here and they are a and bi Up.
[0326] [51 number]
They are the becoming arbitrary complex numbers. Above-mentioned w rho
1 Complex Number from 2 Conversion -- Rectangular Conversion To -- a and B are Intersected Perpendicularly on Complex Plane like -- Namely
[0327] [52 number]
SR[cl *c2]= 0
It is good also as a becoming complex number. Here, * expresses complex conjugate,
[0328] [53 number]
[]
They are Is and an operation which takes out the real part of a complex number, and an imaginary part, respectively.
[0329] In this case, in detection series extraction part 221, it may ask for value [ of detection series 1113 ] i , and gamma using the inverse transform of the above-mentioned conversion from complex number q of detection complex pattern 961.
1 2
[0330] For example, respond to the value of w of one value taken out from embedding series 913, and it is QA.
It is a point (for example) on a complex plane like M abnormal conditions.
[0331] As four points of [a 54 number] are chosen, even if it uses the complex number value of the selected point, it is right .
[0332] Feature > of the embodiment of Mr. 1
There is the feature shown below in this above-mentioned embodiment.
[0333] According to electronic watermark embedding device 100 and digital-watermarking detecting device 200 of a reduction book embodiment of a noise ingredient In the embedding of electronic Power, it is a cycle at time modulation part 130. It sets to detection of electronic Power to digital watermarking modulated and embedded by the signal. It detects by performing integration calculation with a periodic signal by time demodulation section 210. Thereby, is it Electronic penetration? Distribution of the noise ingredient added signal 912 before embedding which serves as a noise for carrying out, and after that It becomes small. (The periodic signal is decided that the value which found the integral by one cycle is set to 0, and depends this on To) It sets especially to a picture signal, The approaching correlation inter-frame [ each ] is comparatively high. Things are known and correlation of the direction of time is quantity within a cycle by a periodic signal and integration calculation. The distribution which it is, and an ingredient is removed, embeds as a result, and originates in last signal 912 is dramatically small. It becomes.
[0334] The above-mentioned literature "Kana Yamamoto, Takao Nakamura, Yoichi Takashima, Jun Katayama, According to Ryo Kitahara and * Taketaka "1 about detection quality assessment of frame superposition type video electronic Power consideration" information-science-and-technology forum F IT2005, J-029, and 2005" Place of electronic Power using spectral diffusion and correlation computation The detection evaluation value showing the reliability of detection in the meaning of the false positivity of Synthesis and electronic Power is embedded. A detection evaluation value becomes large, so that the distributed value of a last signal is small. (It is contained in a part mother's B, and C by the above-mentioned formula (3) and (4).)
[0335] [55 number]
But, so that it is small -- expected value [ of a detection evaluation value ] E[<sub>P</sub>] becomes large -- this -- correlation value calculating part 222 it sets to correlation computation, so that distribution of the noise ingredient contained in gamma is small -- reliable Inspection meaning that appearance becomes possible -- electronic Penetration of the present invention An embedding device and Electronic penetration It lends and expresses that the reliability of detection is high with a detecting device.
[0336] N -- Cycle which generates one-dimensional embedding complex pattern 921 as complex number arrangement based on embedding series 913 made using pseudorandom numbers and from which 90 degrees of phases differ it It is modulating by signal, the phase of embedding pattern 922 -- N -- a one-dimensional space top -- diffusion -- To be -- noise which becomes like, embeds at the time of detection of electronic Power, originates in last signal 912, and appears The size of an ingredient becomes smaller. a result -- the more reliable embedding of electronic Power, and detection becoming possible -- reliability comparable as the former -- Burial of electronic Power with less quality degradation Because -- a lump -- it becomes detectable.
[0337] Spectral diffusion series length increase and N of a complex number -- It is embedding the embedding series of digital watermarking as a one-dimensional pattern, For example, the above-mentioned literature "Takao Nakamura, Hiroshi Ogawa, and Tomioka Jun*, way method "of the improvement in parallel translation' cutoff tolerance in Yoichi Takashima digital watermarking 1999 Be like Year Code, Information Security Symposium, and SCIS99-W3-2.1, Pp. 193-198, and 1999". Are . The Electronic Penetration Method and Comparison Which Repeat Still Picture-oriented Electronic Power and it Embeds on Each Frame It Carries Out, Twice as many spectral diffusion series length as this can be used.
[0338] The above-mentioned literature "Kana Yamamoto, Takao Nakamura, Yoichi Takashima, Jun Katayama, According to Ryo Kitahara and * Taketaka "1 about detection quality assessment of frame superposition type video electronic Power consideration" information-science-and-technology forum F IT2005, J-029, and 2005" In proportion to the square root of the series length of spectral diffusion and place diffusion of electronic Power using correlation computation, it becomes large. Synthesis and the detection evaluation value showing the reliability of detection in the meaning of the false positivity of digital watermarking are spectra. (being (4) the formula (3) of the above-mentioned literature, and E[)<sub>P</sub>If, as for this to which] is proportional to molecule 1, spectral diffusion series length becomes long, It means that detection with the high part reliability is attained, and is an electronic watermark embedding device of the present invention. It reaches, a twice as many detection evaluation value as this is acquired in a digital-watermarking detecting device as compared with a conventional system, and it is Inspection. It expresses that the reliability of appearance is high.
[0339] moreover -- burying by embedding respectively separate information instead of doubling spectral diffusion series length, if it is when the reliability of detection comparable as the former is required as a whole -- Included It can see and can also double information length. [0340] In abnormal-conditions time modulation part 130 to the direction of N dimension, N -- It is a spectrum to one-dimensional space. Diffused N -- One-dimensional embedding pattern, the direction of eye N dimension intersect perpendicularly with it -- circumference N -- it is modulating using a term signal -- as opposed to the synchronous displacement given in the direction of eye N dimension -- have common influence in one-dimensional space! /and To! /-- it obtains and has the feature.
[0341] N -- Redundancy which spread the embedding information on a one-dimensional pattern to N dimension space, For example, in the case of a picture signal, sufficient tolerance is Ah also to change of high compression, re-photography, etc. Ri, When a part of signal (the case of a picture signal, for example, some frames) is changed, Signal When a part is started, it becomes detectable [ * (in It was case / Extracting some frame sections, for example in the case of a picture signal. /) straw matting electronic Power ], and it is Depression about quality degradation also to these cases. It obtains and can perform detection of the merit of information length, information Stickiness , and Electronic penetration.
[0342] In prevention and time modulation part 130 of vulnerable frame generating, embed the sum of the signal which intersects perpendicularly or was modulated by two periodic signals by which phases differ, and use as a pattern. the case where it modulates only by one periodic signal -- case of the embedding to a picture signal all the values of an embedding pattern turn into less than the minimum picture signal quantization value -- actually -- The frame into which the embedding of electronic Power is not made may occur. moreover -- the -- leave a frame [ like ] and the amplitude of the ingredient of Power is actually embedded fully greatly -- and To The attack which deletes electronic Power by changing targeting a frame will also be attained. Book 2 from which it intersects perpendicularly with like the electronic watermark embedding device of an embodiment, or a phase differs It is embedding the sum of the signal modulated by the periodic signal of One, and using as a pattern, the minimum it can prevent The deflection into which it becomes less than a picture signal quantization value, and the embedding of electronic Power is not made in fact - Beam occurring, and effective in the picture signal as a transmission way of electronic Power -- profit While being able to carry out for When it changes targeting a frame with large amplitude of electronic Power, it is Attack when. The receiving tolerance can be increased.
[0343] the effect of detection from difference' differentiation -- again -- as time demodulation section 210 -- a figure -- recovering from a difference value and a differential value, if the composition of 27 is used, more than the offset part by a periodic signal -- original picture ingredient When it is canceled greatly, detection of electronic Power becomes easy and detecting accuracy improves, it is. It obtains and there is an advantage. On the contrary, even if it is the embedding of weaker electronic Power, it is comparable detection performance. Since it can maintain, With solution of a [0344] in which embedding of less electronic Power of degradation of signal is possible synchronous subject, and the digital-watermarking detecting device in this embodiment, Profit With the electronic watermark embedding device in this embodiment although for is not carried out, The digital-watermarking detecting device described by the embodiment of The 4 mentioned below and the embodiment of The 5 is used, N -- The embedding series by which spectral diffusion was carried out in one-dimensional space is a pair to a synchronous gap in the direction of eye N dimension. The needlessness using carrying out and being subject to common influence of matching synchronization, or easy and a high speed It has the feature that possible electronic Power of matching synchronization can be embedded.
[0345] In the electronic Penetration embedding device in the tolerance and this embodiment to the direction scaling of time, As frequency of the periodic signal used by time modulation part 130, it is comparatively low. Electron which has a certain amount of tolerance also, for example to the attack accompanied by elasticity of the directions of time, such as Beam [ Frame rate conversion, frame drops, and deflection - ] insertion, if frequency is used Watermark detection is possible.
[0346] Digital watermarking required [ the whole effect and since the reliability of high Resistance of detection of electronic Power is increasing as a whole ] in order to acquire the reliability and tolerance of detection comparable as the former Since embedded intensity is small and ends, quality degradation of the signal by digital watermarking -- more -- smallness -- To can be carried out -- for example, the case of the embedding to a picture signal -- image containing digital watermarking Image quality can be made high.
[0347] [Embodiment of The 2]
One dimension FFT time abnormal conditions >
An electronic watermark embedding device and digital-watermarking detection A device is explained. [ in / to the following / the embodiment of The 2 ]
[0348] The electronic watermark embedding device and digital-watermarking detecting device of this embodiment are the fruits of The 1. The time abnormal conditions of the electronic watermark embedding device of the form of Application, and a digital-watermarking detecting device, recovery place It is the example which realized Reason by one-dimensional Fourier transform processing.
[0349] Carry out the electronic watermark embedding device of an embodiment in Electronic penetration of the embodiment of The 1, and bury it. It has the same composition as lump device 100, and has the composition from which only time modulation part 130 differs.
[0350] The digital-watermarking detecting device of this embodiment has the same composition as digital-watermarking detecting device 200 of the embodiment of The 1, and has the composition from which only time demodulation section 210 differs.
[0351] Digital watermark embedding device 1-hour modulation part >
Drawing 31 can be set to the electronic watermark embedding device in the embodiment of The 2 of the present invention. The example of composition of a time modulation part is shown. time [ to be shown in the figure ] modulation part 130b -- one-dimensional reverse Fourier -- strange It has Conversion part 134. As for one-dimensional inverse Fourier transform part 134, embedding complex pattern 921 is. It is inputted and embedding pattern 922 is outputted.
[0352] Generation processing of embedding pattern 922 by time modulation part 130b is one-dimensional reverse Fourier. Conversion part 134 carries out in the procedure below and The.
[0353] a figure -- 32 is a flow chart of operation of the time modulation part in the embodiment of The 2 of the present invention.
[0354] P (x) to the position (X, X, ..., X) of step 901 embedding complex pattern 921
1 2 N-1 1
Off to specific frequency [ in / for X, ..., X / the axis (for example, time-axis) of eye N dimension ]
2 N-1
It is concluded that it is a 1 Rie coefficient.
[0355] Step 902 Discrete inverse Fourier transform of the Fourier coefficient of Step 901 is carried out to the axis of eye N dimension, and it is a position (X, X, ..., X).
1 Acquire the one-dimensional series over 2 N-1.
[0356] Step 903 N dimension pattern which makes the one-dimensional series of Step 902 the value of each position is embedded, and it is considered as pattern 922.
[0357] Explain below using a formula concretely.
[0358] Embedding complex pattern 921 It is referred to as rho (chi, chi, ..., chi).
1 2 N-1
[0359] Use P (x, x, ..., x) and it is discrete Fourier coefficient pattern xi (X, X, ...) as follows.
1 2 N-1 1 2
u) [ X and ] constitute.
N-1
[0360] [56 number] (x being chi)<sub>2</sub>-, ', chi<sub>lambda</sub>- _iota and kappa (w = t/-<<)<sub>0</sub>) (81)
<img file="WO2007102403A1_D0034.tif" />0 (K!="O, Facial -- "0)
however, * expresses complex conjugate -- the frequency u was beforehand decided to be, and U -- the number of frequency specimens -- Ah It is considered as To. It is discrete inverse Fourier transform to give the conjugate complex number of P by u=u and u=U--u.
0 0
a result -- obtaining -- having -- a signal -- a real value -- becoming -- a sake -- it is .
[0361] Carry out One(ing), In one dimension dispersion inverse Fourier transform of above-mentioned xi to u, and it is embedding pattern W (X, X).
- - -, X, and t are obtained.
2 N- 1
[0362] Digital watermark detecting device 1-hour demodulation section >
Next, it is One to time demodulation section 210 of digital-watermarking detecting device 200 in the embodiment of The 2. Are and it explains.
[0363] a figure -- 33 shows the example of composition of the time demodulation section in the embodiment of The 2 of the present invention.
[0364] a figure -- it has time demodulation section 210c of 33, and one-dimensional Fourier transform part 214, and embedding finishing signal 923 is inputted, and output detection complex pattern 961.
[0365] in addition -- a figure -- in 33 -- a figure -- in order to make correspondence with 31 easy, composition was noticed about a statement now , and To that information flows upwards from the bottom -- !,
[0366] the procedure of the following [ processing / of embedding finishing signal 923 by time demodulation section 210c / recovery ] -- operation -- To be.
[0367] a figure -- 34 is a flow chart of operation of the time demodulation section in the embodiment of The 2 of the present invention.
[0368] Step 1001 Predetermined section T is taken out from inputted embedding finishing signal 923.
[0369] Step 1002 It is one-dimensional dispersion in every position (X, X, - - -, X) about section T of Step 1001.
1 2 N- 1
Frequency decomposition is Fourier-transformed and carried out.
[0370] Step 1003 From the result of Step 1002, the Fourier coefficient of predetermined frequency is taken out and it is considered as detection complex pattern 961.
[0371] Explain below using a formula concretely.
[0372] Embedding finishing signal 923 It is referred to as gamma (chi, chi, - .., chi, t).
1 2 N- 1
[0373] Carry out the one-dimensional dispersion Fourier transform of gamma (chi, x, - .., x, t) as follows, and it is r? (x, x, - .., x)
1 2 N- 1 1 2 N
u is obtained.
[0374] [57 number]<img file="WO2007102403A1_D0035.tif" />
However, T presupposes that it is the predetermined number of specimens decided beforehand.
[0375] Detection complex pattern 961 shall be denoted by Q (x, chi, ..., chi), and it is tau? (chi, chi, ..., chi)
1 2 N-l 1 2]
u power Q (x, chi, ..., chi)
1 2 N-l
Q (x, chi, ..., chi) = 77 (chi, chi, ..., chi, u) (83)
It comes out and asks. However, U presupposes that it is the frequency decided beforehand.
0
[0376] Here explains the time recovery using difference' differentiation.
[0377] As the embodiment of The 1 showed, embed instead of the above and it is ending signal 923 gamma.
The one-dimensional dispersion Fourier transform of the difference or differentiation of the direction of t axis of (X, X, ..., X, and t) is carried out, and it is Recovery.
1 2 N-1
To tune -- it may make . Namely, when difference is used, for example,
[0378] [58 number]
4)<img file="WO2007102403A1_D0036.tif" />
However, T presupposes that it is the predetermined number of specimens decided beforehand. Specimen interval predetermined in At It is. For example, the case of At=l, computing difference at intervals of 1 specimen is shown, and it is a picture signal. When it takes for an example, it means calculating adjoining inter-frame difference. At is by numbers other than one. You may be.
[0379] the example of composition of such a time demodulation section 210 -- a figure -- it is shown in 35.
[0380] a figure -- 210 d of time demodulation sections of 35 -- a figure -- after having the almost same composition as time demodulation section 210c of 33 and inputting embedding finishing signal 923 into - A signal differentiation part 215 One-dimensional Furi It differs in that it is constituted so that it may be inputted into A conversion part 216. [0381] as opposed to embedding finishing signal 923 inputted in signal differentiation part 215 -- the axis of eye N dimension, and To the difference or differentiation of What is and the direction of t axis is calculated -- it outputs to one-dimensional Fourier transform part 216 It is made like.
[0382] It is 1 about the difference or differentiation of the direction of t axis of embedding finishing signal 923 gamma (chi, chi, - .., x, t).
1 2 N- 1
the effect of carrying out a dimension dispersion Fourier transform and making it get over -- the embodiment of The 1 -- the figure to kick -- it is the same as that of the effect in time demodulation section 210b of 27.
[0383] Feature > of the embodiment of Mr. 2
The feature of this embodiment is explained.
[0384] An electronic watermark embedding device and a digital-watermarking detecting device of this embodiment, Electron which acquires the same effect as the electronic watermark embedding device of the embodiment of The 1, and a digital-watermarking detecting device A one-dimensional Fourier transform is used for a watermark embedding device and a digital-watermarking detecting device, and it is a fruit. It is the expressed example.
[0385] Use the existing Fourier transform device by using a one-dimensional Fourier transform, and it is electronic Penetration easily. An embedding device and a digital-watermarking detecting device can be constituted.
[0386] In addition, it is the electronic watermark embedding device and Electronic penetration of the embodiment of The 1, and this embodiment. It may lend and you may use it combining a detecting device. Namely, digital-watermarking embedding Let time modulation part 130b of a device be time modulation part 130a of the embodiment of The 1, digital watermarking Combine time demodulation section 210c of a detecting device as 210d of time demodulation sections of this embodiment. even if it carries out Use -- right -- It carries out and is this embodiment about time modulation part 130a of an electronic watermark embedding device. It is considered as time modulation part 130b which can be set, and is this operation about time demodulation section 210c of a digital-watermarking detecting device. It is good also as 210d of time demodulation sections in a form.
[0387] [Embodiment of The 3]
<Two-dimensional FFT coefficient embedding>
An electronic watermark embedding device and an electron A watermark detecting device is explained. [ in / to the following / the embodiment of The 3 of the present invention ]
[0388] This embodiment is the electronic watermark embedding device and digital watermarking of an embodiment of The 1. In a detecting device, it is electronic Penetration. It is the example which was made to perform embedding in a rectangular conversion field.
[0389] the electronic watermark embedding device of this embodiment -- Digital watermarking of the embodiment of The 1 Because -- although it is the same composition as a lump device, only complex pattern generation parts differ.
[0390] Although the digital-watermarking detecting device of this embodiment is the same composition as the digital-watermarking detecting device of the embodiment of The 1, only detection information extraction parts differ.
[0391] Digital watermark embedding device-complex pattern generation part >
Below, the complex pattern generation part in this embodiment is explained.
[0392] a figure -- 36 shows the composition of the complex pattern generation part in the embodiment of The 3 of the present invention.
[0393] Complex pattern generation part 110b shown in the figure, Embedding series generation part 111, complex arrangement generation part 112, and N -- It comprises one-dimensional inverse Fourier transform part 113, a figure -- the composition of 12 -- N -- one dimension are the composition that the inverse Fourier transform part 113 was added, and embedding information 911 is inputted -- Fill -- lump complex pattern 921 is outputted.
[0394] Generation processing of the embedding complex pattern by complex pattern generation part 110b is carried out in a procedure below.
[0395] a figure -- 37 is a flow chart of operation of the complex pattern generation part in the embodiment of The 3 of the present invention.
[0396] Generate embedding series 913 which is a numerical sequence showing embedding information in step 1101 embedding series generation part 111 based on inputted embedding information 911. Operation of embedding series generation part 111 is the same as that of the embodiment of The 1.
[0397] Step 1102 It is N about embedding series 913 generated by embedding series generation part 112 in complex arrangement generation part 112. -- It divides into the real part and imaginary part of an element on one-dimensional complex arrangement. It guesses and generates middle complex pattern 904.
[0398] Mention operation of complex arrangement generation part 112 below.
[0399] Step 1103 N -- In one-dimensional inverse Fourier transform part 113, as opposed to middle complex pattern 904 generated by complex arrangement generation part 112 -- N -- performing one-dimensional inverse Fourier transform -- said -- QuietlyN -- one-dimensional embedding complex pattern 921 is generated. N -- One-dimensional inverse Fourier transform About mentions below in the details of operation of part 113.
[0400] Electronic power and embed and it is device-complex pattern generation part-complex arrangement generation part >.
Next, it is To explain in detail about operation of complex arrangement generation part 112 of the above-mentioned step 1102. To. [0401] Operation of complex arrangement generation part 112 is the same as that of complex arrangement generation part 112 of the embodiment of The 1. Although it may be operation, it is a place as follows because of the embedding of more effective electronic Power. Reason may be carried out.
[0402] 1 -- Size M X M X which is Value power^ of all the element -- N of X M -- one-dimensional Complex number distribution
1 2 N- 1
A sequence is prepared. However, M, M, - .., and M are the numbers of elements defined beforehand.
1 2 N- 1
[0403] Determine the range of the element which should embed in 2 complex arrangement and should assign series 913. The example of the range is mentioned below.
[0404] The values them 3 embedding series 913 power also took out two values at a time in order, and took out in order to the element within the limits defined by above 2 in the above-mentioned arrangement of one are a real part of the element value, and an imaginary part. A value is set to arrangement like.
[0405] Use as middle complex pattern 904 complex arrangement generated by 4 above 3, and it is N. -- It outputs to one-dimensional inverse Fourier transform part 113.
[0406] It is also until it changes an order of embedding series 913 to a random order using pseudorandom numbers and right and a thing also say VGo in advance of above 3 like complex arrangement generation part 112 in the embodiment of The 1.
[0407] Processing of such complex arrangement generation is shown in patent documents 1, and is similar to generation of and a Watermark coefficient matrix. However, in the present invention, middle complex pattern 904 is N behind. -- 1st order Although inverse Fourier transform is carried out in former inverse Fourier transform part 113, The result by which inverse Fourier transform was carried out is a fruit. As for the necessity of becoming a numerical value, since it may become a complex number dropping off, middle complex pattern 904 is Furi. It is not necessary to hold the symmetry of a A conversion factor. Namely, all the above-mentioned ranges of 2 It can embed using an element, embedding of series 913 can be performed, and it compares with patent documents 1. It means burying and putting The and embedding series 913 of twice as many length as this.
[0408] Explain the range of the element in above 2.
[0409] Figure 38A*38F can be set to the complex arrangement generation part in the embodiment of The 3 of the present invention. It is each example of the element range of complex arrangement. Figure It can set to 38A, B, and C at complex arrangement generation part 112. The example of the range of the element of a complex array is shown. N = be for an example about the case where two-dimensional complex arrangement is used by 3. It is shown. Figure In 38A*38F, a square portion expresses complex arrangement, and it is credit mosquito Ivy of a net. The range of the element which a field should embed and should assign series 913 is shown. [0410] complex arrangement -- after -- N -- inverse Fourier transform is carried out in one-dimensional inverse Fourier transform part 113 , -- it is possible that it is expression in the frequency domain to an embedding pattern.
[0411] since the element (0, 0) of an array expresses DC ingredient now -- a figure -- rewriting complex arrangement in round like 38D, E, and F, so that the element (0, 0) of an array may come to a center, embedding series 913 power S -- he can understand using what kind of frequency band it is embedded.
[0412] a figure -- 38D -- a square area and a figure -- 38E -- a circular field and a figure -- 38F -- a lozenge field -- any It becomes the embedding to an inside frequency band.
[0413] for example, the case where the digital-watermarking embedding to a picture or an image is considered -- the number zone of high frequency -- Fill -- Included -- rare -- Electronic power image numerals Y, such as MPEG 2 and H.264, -- easy -- deletion -- While Have been striped A result can be considered, in size, influence with visual embedding to a low frequency wave field comes. It was and mentioned the embedding from things to an inside frequency band as the example.
[0414] moreover -- comparing the frequency of the direction of slant with the frequency of the direction in every direction in a high frequency region on man's vision characteristic -- vision sensitivity -- low -- Even if things are known, it is ("* original sincerity "systematic image coding" and eye PC, 1990, pp.87"), it compares that it is in every direction and it performs embedding to a low frequency band There is the characteristic of being hard to be conspicuous. pictures, such as JPEG and MPEG, and image numerals Y etc. -- tendency for high frequency to tends to be reduced by numerals Y comparing the direction of the direction of slant with the direction in every direction since the quantization step of the direction of slant is set up more greatly than the direction in every direction It is. Figure Such [ performing embedding to a lozenge field like 38C and F ] a situation smell Also in The, picture quality can perform electronic Penetration embedding with high high Resistance.
[0415] a complex arrangement generation part [ in / in figure 39A*39F / the embodiment of The 3 of the present invention ] -- It is each example of the element range of the complex arrangement to kick. using the prolonged signal of the direction of time, i.e., many frame numbers, like the present invention in the case of the digital-watermarking embedding to an image -- Inspection the case where digital-watermarking detection with high appearance performance is performed -- the intensity of the embedding of electronic Power -- smallness -- To can be carried out. as a result, the number field of low frequency waves as shown in figure 39A*39F -- burying -- Included Even if it performs Body, can make visual influence small, and take such a method of embedding. It is possible.
Electronic power(ing) -- Embedding device-complex pattern generation part N -- N of one-dimensional inverse Fourier transform part >, next complex pattern generation part 110b -- operation of one-dimensional inverse Fourier transform part 113 -- just -- The -- it explains in detail.
[0416] N -- The Fourier transform in one-dimensional inverse Fourier transform part 113 is performed by the following procedures.
[0417] N generated by 1 complex arrangement generation part 112 -- One-dimensional middle complex pattern 904<img file="WO2007102403A1_D0037.tif" />
It carries out.
[0418] One-dimensional ?^carries out discrete inverse Fourier transform of the 2 alpha (chi, X, - - -, X), and it is embedding complex butter.
1 2 N- 1
It asks for - A P (x, X, --, X).
1 2 N- 1
[0419] [59 number]
A 1-]
.
<img file="WO2007102403A1_D0038.tif" />
(8 6)
However, mu, mu, - .., and mu were generated by complex arrangement generation part 112. nu -- One-dimensional middle
1 2 N- 1
It is a size of complex pattern 904 (the number of elements of each dimension).
[0420] Digital watermark detecting device Detection information extraction part >
Next, detection Information extraction of digital-watermarking detecting device 200 in the embodiment of The 3 of the present invention Appearance part 220b is explained.
[0421] a figure -- 40 shows the example of composition of the detection information extraction part in the embodiment of The 3 of the present invention.
[0422] Detection information processing section 220b shown in the figure, N -- One-dimensional Fourier transform part 225, detection series From extraction part 221, correlation value calculating part 222, maximum judgment part 223, and detection information reconstruction section 224 It is constituted, detection complex pattern 961 is inputted, and detection information 914 is outputted.
[0423] in addition -- a figure -- in 40 -- a figure -- which composition was indicated that information flows upwards from the bottom, and was noticed about and To in order to make an understanding of correspondence with 36 easy,
[0424] Detection information extraction processing by detection information extraction part 220b is carried out in the following procedures.
[0425] a figure -- 41 is a flow chart of operation of the detection information extraction part in the embodiment of The 3 of the present invention. [0426] detection complex The inputted in step 1201 N--l dimension Fourier transform part 225 N -- as opposed to turn 961 -- N -- performing a one-dimensional Fourier transform -- the same -- a one-dimensional detection complex Arrangement 1115 is generated.
[0427] N -- The details of operation of one-dimensional Fourier transform part 225 are mentioned below.
[0428] it sets to step 1202 detection series extraction part 221 -- N -- the value of the real part of an element value and the imaginary part was taken out and put in order from detection complex arrangement 1115 generated in one-dimensional Fourier transform part 225 Detection series 1113 is constituted.
[0429] Mention the details of operation of detection series extraction part 221 below.
[0430] Step 1203 Detection series 1113 which comprised detection series extraction part 221 in correlation value calculating part 222, and embedding series constituted based on the embedding series assumed Correlation is calculated and correlation value 1114 is calculated.
[0431] Operation of correlation value calculating part 222 is the same as that of the embodiment of The 1.
[0432] Step 1204 Correlation acquired by correlation value calculating part 222 in maximum judgment part 223 Correlation value calculation corresponding to correlation value 1114 which value 1114 finds the thing used as the maximum, and serves as the maximum The embedding series used by the correlation computation in part 222 is determined.
[0433] Operation of maximum judgment part 223 is the same as that of the embodiment of The 1. electronic Penetration burying -- Included It sees and is based on maximum judgment part 223 depending on the constitution method of the embedding series in device 100. it is made to be judged by other methods instead of a maximum judging -- Go -- right -- until things say There is nothing.
[0434] Step 1205 It is a re-style about detection information 914 it is judged that it was actually embedded in detection information reconstruction section 224 based on the embedding series judged by maximum judgment part 223. Formation is carried out.
[0435] Operation of detection information reconstruction section 224 is the same as that of the embodiment of The 1.
[0436] Digital watermark detecting device Detection information extraction part N -- One-dimensional inverse Fourier transform part >
Next, N of the above-mentioned detection information extraction part 220b -- Operation of one-dimensional Fourier transform part 225 is full. Fine is explained.
[0437] Fourier transform processing in N- one-dimensional Fourier transform part 225 is performed by the following procedures. To. [0438] From detection complex pattern 961 by which 1 input was carried out to a size M XM chi...chimu nu -- One dimension
1 2 N-1
Of complex arrangement is constituted. However, mu, mu, ..., mu are electronic watermark embedding devices 100.
1 2 N-1
It is the number of elements same with having used by Of complex arrangement generation part 112.
[0439] When detection complex pattern 961 is obtained as a discrete signal, it is N as it is about it. -- It is considered that it is one-dimensional complex arrangement. Detection complex pattern 961 is obtained as a continuation signal. What carried out specimen Y of the detection complex pattern 961 using arbitrary specimen Means when it was
nu -- It uses as one-dimensional complex arrangement.
[0440] When setting complex arrangement of 1 of 2 above to Q (xX, ..., X), it is N about this. -- One-dimensional Separation
1 2 N-1
A Scatter Fourier transform is carried out and it asks for A [u, u, ..., u].
1 2 N-1
[0441] [60 number]
--,<sup>M</sup>,<sup>l Mn</sup>-,<img file="WO2007102403A1_D0039.tif" />
(87)
3) Consider A [u, u, ..., u] for which it asked by above 2 as detection complex arrangement 1115, and it is a detection series.
1 2 N-1
It outputs to extraction part 221.
[0442] Digital watermark detecting device Detection information extraction part Detection series extraction part >
Next, the details of operation of detection series extraction part 221 of detection information extraction part 220b are explained. It carries out.
[0443] Although operation of detection series extraction part 221 is fundamentally [ as the embodiment of The 1 ] the same, even if it unites with operation of complex arrangement generation part 112 in this embodiment and processes as follows It is good.
[0444] Processing by detection series extraction part 221 is carried out in the following procedures.
[0445] 1 N -- From detection complex arrangement 1115 obtained in one-dimensional Fourier transform part 225, Electronic penetration The element of the range which was carried out and was used by complex arrangement generation part 112 of embedding device 100 is taken out and taken, and it comes out. The real part of the complex number value carried out and an imaginary part are seen as an independent real value, respectively, and are put in order. This is Electricity. Child permeability is carried out and it is as symmetrical as generation of the complex arrangement by complex arrangement generation part 112 of embedding device 100. It is processing.
[0446] Two The acquired series is set to i , gamma, ..., gamma, and it outputs as detection series 1113.
1 2 L
[0447] When exchange of an order of embedding series 913 or the element of a complex array is made in complex arrangement generation part 112 of electronic watermark embedding device 100, it is Return about the order. Point is the same as that of a 1st embodiment.
[0448] Feature > of the embodiment of Mr. 3
Below, the feature of this embodiment is described.
[0449] According to the electronic watermark embedding device and digital-watermarking detecting device of this embodiment example what is shown with For example and patent documents 1 -- the same -- the numerals I noise of a picture signal -- improper -- remaining, even if it sets -- Easy and -- embedding electronic Power in a zone with little visual influence since it can do -- electronic Penetration of high Image quality with high tolerance Embedding is possible.
[0450] Robust electronic Penetration embedding is possible by the compression in image numerals Y etc. especially by performing the range of the embedding in a frequency domain in a lozenge field.
[0451] If an offset searching method as shown with patent documents 1 is taken since the embedding pattern is diffused to the whole signal, digital-watermarking detection is possible for it also from the signal partially started from the embedding finishing signal.
[0452] Electronic Penetration furthermore shown with patent documents 1 by the present invention Differing from a method, middle complex No turn 904 is, the watermark indicated to paragraph number 0197 of patent documents 1 -- coefficient matrix Since it is not necessary to maintain the symmetry of a Fourier transform coefficient as compared with When, twice as many length as this buries. Lump series 913 can be embedded. namely, twice as many spectral diffusion series length as this -- business it comes out that it is -- last .
[0453] If spectral diffusion series length becomes long as already stated, If detection with the high part reliability is attained and it is the reliability of detection comparable as the former, If embedding information length can be doubled and it is detection reliability comparable as the former, and information length, it is Low quality more. Electronic Penetration with little-izing Since embedding becomes possible, By the present invention, reliability is high Regret. Electronic Penetration with little news length to long Quality degradation It is shown that embedding becomes possible.
[0454] Time modulation part 130b or time demodulation section 2 of this embodiment and the embodiment of The 2 It is also possible to carry out combining 10b.
[0455] Time modulation part 130 of this embodiment and the embodiment of The 2 Or it is N when combining time demodulation section 21 Ob. -- N of one-dimensional inverse Fourier transform part 113 -- One-dimensional reverse Off A 1 Rie conversion process and one-dimensional inverse Fourier transform processing of time modulation part 130 are put in block, and it is N dimension. It can also carry out as inverse Fourier transform processing. Similarly, it is N. -- One-dimensional Fourier transform N of part 225 -- A one-dimensional conversion process and one-dimensional Fourier transform processing of time demodulation section 210 are put in block. It can carry out and can also carry out as Fourier transform processing of N dimension.
[0456] [Embodiment of The 4]
Time synchronous unnecessary detection >
An electronic watermark embedding device and digital-watermarking detection A device is explained. [ in / to the following / the embodiment of The 4 ]
[0457] This embodiment is set to a digital-watermarking detecting device, when embedding of digital watermarking is performed using electronic watermark embedding device 100 in the embodiment of The 1, Time It is synchronous Fit when the signal with which the synchronization shifted in the direction of an axis (axis of the Nth dimension) is inputted. It is an example for performing digital-watermarking detection without the necessity of performing To.
[0458] The digital-watermarking detecting device of this embodiment is digital-watermarking Detector of the embodiment of The 1. It has the same composition as Place 200, and has the composition from which only detection information extraction part 220 differs.
[0459] About time demodulation section 210, the time demodulation section of the embodiment of others of the present invention may be used. For example, it is a style even if it uses time demodulation section 210b of the embodiment of The 2, and 210c. Do not know.
[0460] In addition at this embodiment, it is, electronic watermark embedding device 10 in the embodiment of The 1 Power in which the case where embedding of electronic Power is performed using 0 is indicated for the example others -- operation Embedding of electronic Power is performed using the electronic Penetration embedding device in a form. Also detection of a It was case It can combine similarly and can apply. For example, The 3 The electronic watermark embedding device in an embodiment performs embedding, and it is a form of operation of The 3. N of the digital-watermarking detecting device in voice -- One-dimensional Fourier transform part 225 is combined, and it is Inspection. It may be made to perform appearance. those cases -- case where a suitable change is required for each procedure although it is -- those required change -- explanation of this embodiment, and Burial of those digital watermarking Because -- if based on explanation of lump Stickiness and The, until it will say a clear thing --
[0461] Digital watermark detecting device Detection information extraction part >
Operation of a detection information extraction part [ in / to the following / the digital-watermarking detecting device of this embodiment ] It explains.
[0462] a figure -- 42 shows the composition of the detection information extraction part in the embodiment of The 4 of the present invention. in the figure -- a figure -- 28 and a figure -- the same numerals are given to 40 and an identical configuration portion.
[0463] Detection information extraction part 220c shown in the figure, it comprises detection series extraction part 221, complex correlation value calculating part 226, absolute value calculation part 227, maximum judgment part 223, and detection information reconstruction section 224 -- detection complex pattern 961 is inputted and detection information 914 is outputted.
[0464] Detection information extraction processing by detection information extraction part 220c is carried out in the following procedures.
[0465] a figure -- 43 is a flow chart of operation of the detection information extraction part in the embodiment of The 4 of the present invention.
[0466] Step 1301 Detection complex pattern 96 inputted in detection series extraction part 221
Detection complex number series 1118 which put in order the complex number value acquired from 1 is constituted. Detection series extraction Detailed operation of part 221 is mentioned below.
[0467] Step 1302 In complex correlation value calculating part 226, detection series extraction part 221 -- composition -- Complex [ which was constituted based on Was done detection complex number series 1118 and the embedding series assumed ] The complex correlation with a number series is calculated and complex correlation value 1116 denoted by a complex number is calculated.
[0468] It is it about the complex correlation with a plurality of complex number series constituted based on a plurality of embedding series considered when the value which changes with kinds of embedding series was embedded. But calculation is done and corresponding complex correlation value 1116 is calculated.
[0469] Mention the details of operation of complex correlation value calculating part 226 below.
[0470] Step 1303 In absolute value calculation part 227, it was obtained by complex correlation value calculating part 226. Absolute value 1117 of complex correlation value 1116 is computed. By computing an absolute value, it is a gap of a synchronization. Also when Signal is inputted, it is not concerned with the amount of synchronous displacement, but it is the quantity of correlation with a detection complex number series.
V and an embedding series can be determined.
[0471] in step 1304 maximum judgment part 223, it was obtained by absolute value calculation part 227 -- absolute Complex correlation value calculation corresponding to the absolute value which value 1117 finds the thing used as the maximum, and serves as the maximum The embedding series used by the correlation computation in part 226 is determined.
[0472] In addition, be made to be judged by other methods instead of the maximum judging by and maximum judgment part 223 depending on the constitution method of the embedding series in electronic Penetration embedding device 100. Me me is also good.
[0473] the details of operation of maximum judgment part 223, and the details of other methods used as substitution -- back -- Description is carried out.
[0474] detection information 914 it is judged that it was actually embedded in step 1305 detection information reconstruction section 224 based on the embedding series determined by maximum judgment part 223 -- re--- It constitutes.
[0475] Operation of detection information reconstruction section 224 is the same as that of the embodiment of The 1.
[0476] In addition, it is the embedding power of electronic Power. Digital-watermarking embedding in the embodiment of The 3 When carried out using device 100, Detection series extraction part 221 is preceded and it is operation of The 3. N in a form -- The same Fourier transform processing as one-dimensional Fourier transform part 225 is needed.
[0477] After embedding by maximum judgment part 223 and determining a series, The signal as for which and an input signal displaced the synchronization one by one using the embedding series is used, Detection in the embodiment of The 1 The same processing as information extraction part 220a is performed in round robin, the amount of synchronous displacement is measured, and it is detection information. It is based on the correlation value acquired by extraction part 220, and is To go about acquisition of a more exact detection phase Seki evaluation value. The is not cared about, either. even in this case -- all the embedding series considered -- all the synchronizations -- strange about -- becoming detectable at high speed far cannot be overemphasized than searching in round robin
[0478] Digital watermark detecting device Detection information extraction part Detection series extraction part >
next, the details of operation of detection series extraction part 221 of the above-mentioned detection information extraction part 220c -- just -- Explanation is carried out.
[0479] Detection series extraction part 221 constitutes detection complex number series 1118 from a complex number value acquired from detection complex pattern 961.
[0480] Processing by detection series extraction part 221 is carried out in the following procedures.
[0481] One From detection complex pattern 961 to a size M X M chi - - - chi mu N -- One-dimensional Complex distribution A sequence is constituted. a figure [ in / in the method of composition / the embodiment of above-mentioned The 1 ] -- Step 80 of 30 It is the same as that of processing of 1.
[0482] Two Every one complex number value is taken out and arranged sequentially from the complex arrangement obtained by above 1, and this is made into detection complex number series 1118. That is, complex arrangement is A [p, p, ..., p] (p).
1 2 N- 1 n
* 0 and detection complex number series 1118 gamma, i", - - -, when it is expressed i",
1 2 N- 1
= A [0, 0, ..., 0] (88)
= A [1, 0, ..., 0]
3) Acquired series gamma, gamma, ..., gamma are outputted as detection series 1115.
1 2 L
[0483] It is the order when exchange of an order of embedding series 913 or the element of a complex array is made in complex arrangement generation part 112 of electronic watermark embedding device 100. The point to return is the same as that of detection series extraction part 221 in the embodiment of The 1.
[0484] In addition, it is electronic Penetration embedding power. When carried out using electronic watermark embedding device 100 in the embodiment of The 3, In the above-mentioned processing of 2, it is an embodiment of The 3. Element within the limits used by complex arrangement generation part 112 of electronic Penetration embedding device 100 which can be set A complex number value is taken out and it is made to arrange.
[0485] Digital watermark detecting device Detection information extraction part-complex correlation value calculating part >
Next, it is One in detail in operation of complex correlation value calculating part 226 of the above-mentioned detection information extraction part 220c. Are and it explains.
[0486] Complex correlation value calculation processing in complex correlation value calculating part 226 is carried out in the following procedures. To.
[0487] One They are and embedding series w considered by the same procedure as embedding series generation part 111 of electronic Penetration embedding device 100.<sup>(1)</sup>, w<sup>(2)</sup>-- is generated. The method of generation is the same as that of correlation value calculating part 222 in the embodiment of above-mentioned The 1.
[0488] Two Embedding series w acquired by above 1<sup>(1)</sup>, w<sup>(2)</sup>carrying out electronic Penetration and embedding from --, -- Wear complex arrangement generation part 112 of Place 100 -- the real part and imaginary part of and the complex number value of one The -- assignment Complex number series which makes a Was done value a group, constitutes the same complex number sequence, and embeds this It is considered as xi xi and --. namely, a figure -- complex arrangement is constituted like 17 -- if it becomes -- [a 61 number]<img file="WO2007102403A1_D0040.tif" />
When referred to as (L' =L/2), [ however, ]
[0490] [62 number]
(8 9}
'<sup>= W</sup>i 1 + A<sup>W</sup>i
It carries out.
[0491] A constitution method of an embedding complex number series, in what is limited to this example, it corresponds to the composition of complex arrangement from complex arrangement generation part 112 of electronic watermark embedding device 100 dropping off -- as -- style Accomplish! /If -- even if it is a constitution method [ how ], until style trap and a thing say! /
[0492] Detection series 1115 acquired by 3 detection series extraction part 221, and each embedding series 6 acquired by above 2<sup>(2)</sup>-- Correlation is calculated using complex correlation, respectively.
[0493] Perform correlation computation as follows.<sub>rho</sub><sup>omega</sup>If it is I would like to ask complex correlation value 1116,
[0494] [63 number]
<img file="WO2007102403A1_D0041.tif" />
6 expresses here the sequence of numbers which the conjugate complex number power of the element of 6 R also becomes, and 6 k is 6 k.<sup>omega</sup>Conjugate Suppose that it is a complex number. '" expresses an inner product operation when a sequence of numbers is regarded as a vector. [0495] Here, ^ becomes a complex number.
[0496] above-mentioned literature "Takao Nakamura, Jun Katayama, Masashi Yamamuro, and the Sonehara Ascent -- "high-speed digital-watermarking detection method of analog picture power using a camera cell phone machine", In order to arrange the valuation basis of IEICE TRANSACTIONS D-II, Vol. J87-D -II, No.12, pp. 2145-2155, and detection reliability that is in 2004", For example, regular Y of each element of gamma and ^ is carried out so that the absolute value of average 0 and distribution may be beforehand set to 1. It is an embodiment of The 1 that it may set and may calculate by multiplying by an absolute term by correlation value calculation. It is the same as that of correlation value calculating part 222 which can be set.
[0497] Explain below that detection of electronic Power is possible to the input from which the synchronization shifted by such an operation.
[0498] The series embedded with the electronic watermark embedding device is w= now. It is {w, w, - .., and w}, and is This.
1 What Arranged 2 L Re in Complex Number Xi = 1, Xi, - - -, Xi
2 Carry out and consider it as '}.
[0499] By the embedding of electronic Power, the signal before embedding and other noise signals are a complex series of After joining i' = {i, i, - .., i
1 If [ Take Two and ] '},
i' =i+xi (91)
Series acquired after synchronous displacement of the direction of time is given to this If gamma,<img file="WO2007102403A1_D0042.tif" />
By the above-mentioned formula If correlation with xi is calculated,
[0500] [64 number]
= (i +<sup>jAe</sup>- xi*
<img file="WO2007102403A1_D0043.tif" />
(V V
A0
k + J k<img file="WO2007102403A1_D0044.tif" />
i If xi is independent and L' is large enough,
[0501] The expected value of [a 65 number] is 0,
[0502] [66 number]
<img file="WO2007102403A1_D0045.tif" />
It depends,
[0503] [67 number]
(9 5)
k=l
When electronic Power is not embedded on the other hand,
[0504] [68 number]
<img file="WO2007102403A1_D0046.tif" />
Term since it comes out, and it is and this expected value is 0, in case digital watermarking is embedded as for I p I Wait value
[0505] [69 number]<img file="WO2007102403A1_D0047.tif" />
It compares and detection of small To say electronic Power is possible enough.
[0506] Digital watermark detecting device Detection information extraction part Maximum judgment part >
Next, maximum judgment part 223 of detection information extraction part 220c is explained in detail.
[0507] Processing in maximum judgment part 223 is carried out in the following procedures.
[0508] One Absolute value 1117 obtained by absolute value calculation part 227
I (1) I I (2) I
I /O I and I p I ...
Absolute value I from which Or et al. and a value serve as the maximum<sup>omega</sup>I is found.
[0509] I I =MAX(I p<sup>(1)</sup>I and I rho<sup>(2)</sup>I , (97)
However, mualphachiomicron is an operation which returns the maximum.
[0510] Embedding series w corresponding to 2 I plate I<sup>max)</sup>To obtain.
[0511] The greatest correlation value I p<sup>omega</sup>It judges whether I is over the predetermined threshold, and is predetermined. A threshold is exceeded. Electronic Power is embedded at and a case, and it is judged as and a It is a force thing. It may be made like.
[0512] below, boil, attach and explain operation used as substitution of maximum judgment part 223.
[0513] Instead of the maximum judging by maximum judgment part 223, they are all the embedding series w at complex correlation value calculating part 226.<sup>(1)</sup>, w<sup>(2)</sup>It is embedding series w, without calculating complex correlation to the complex series corresponding to --.<sup>(1)</sup>Complex number series which is alike and corresponds xi<sup>(1)</sup>Complex correlation is calculated to In order and it is obtained. The absolute value of the Complex correlation value judged whether it would be over the predetermined threshold, and exceeded the threshold. It is w about an embedding series.<sup>(max)</sup>It is good, even if it carries out and correlation computation is ended at the time. It is and is ..
[0514] Embedding series power in electronic watermark embedding device 100 For example, as [ showed / (Example 1) of embedding series generation part 111 ] -- comprising only one kind of embedding series -- Fill -- Included -- rare -- The -- !. / To case 2, for example, the (example) of embedding series generation part 111, and (Example 4) showed. Like and since only one is calculated as for and a complex correlation value when it comprises a difference in the positive/negative of one kind of embedding series and is embedded, maximum judging by maximum judgment part 223 a meaning -- The absolute value obtained instead judges by whether they are and To exceeding a predetermined threshold. It may be made to be carried out.
[0515] It spaces in the size of the absolute value of a complex correlation value, and may make it evaluate the reliability of detection.
[0516] Embedding series power in electronic watermark embedding device 100 For example, As [ showed / (Example 2) of embedding series generation part 111 and (Example 4) ] It is Bit by the difference in the positive/negative of an embedding series. It is a synchronous gap, when being constituted and embedded so that OZ1 of a A value may be expressed. The information which the case where the signal with which half a wave of intermediary phase shifted is inputted, and all the bit values reversed Distinction with the case where it is embedded does not stick. In such a case, for example, bit value It embeds so that 1 (or 0) may certainly be taken by making inner 1 bit into the bit for a judgment, and a series is constituted. It sets and may make it amend bit inversion using the bit value. asymmetrical -- incorrect -- It is good also as the judgment by carrying out numerals Y using Correction numerals being possible. Electronic penetration of the present invention It may be made to judge using a different digital-watermarking signal from Power. It is Limit to these examples. Until it says that bit inversion may be amended by the method of others which drop off in the constant thing to carry out There is nothing.
[0517] When being spread in the polarity of positive/negative for every bit using the partial sequence of an embedding series as shown in (Example 4) of embedding series generation part 111, mouth bust detection processing can be performed as follows.
[0518] The partial complex series corresponding to the a-th bit position in detection complex number sequence 1118
[0519] [70 number]
Mo) - j>>{a \
<sup>1</sup>1 T 2, --, T M A
It carries out. It is the complex embedding series used for diffusion of the bit value of the a-th bit position [0520] [71 number]<img file="WO2007102403A1_D0048.tif" />
It carries out. And it is a complex correlation value of each bit position a of every as follows. lambda<sup>a)</sup>It computes.
[0521] [72 number]
.) -- = (.) -<sub>=</sub>^ " -- (.) . * (alpha = \ .. .eta)
Next, eta complex Correlation value<sup>(a)</sup>Direction is arranged. Specifically, it is 0*Arg, for example.<sup>(a)</sup>It becomes To pi. lambda<sup>(a)</sup>it does not change by it being alike and attaching -- pi*Arg<sup>(a)</sup>It is set to To 2 pi. lambda<sup>(a)</sup>It is alike, it attaches and is e.<sup>j</sup>The Multiplied by angle of deviation is rotated 180 degrees. By this change processing, they are all. lambda<sup>(a)</sup>On Is the complex plane It comes to take the value in the first and the second quadrant. How to arrange a direction is Limit et al. to this example. There is nothing. For example, the value in the first on a complex plane and the fourth quadrant can be taken. That is clear.
[0522] Complex Correlation value to which the above-mentioned change processing was performed next<sup>(a)</sup>The sum of It asks for lambda as follows. To.
[0523] [73 number]
And it is in the starting point on a complex plane. A complex plane is divided into two fields by making into a boundary line the straight line which intersects perpendicularly with Arg lambda, and it is before the above-mentioned change processing. lambda<sup>(a)</sup>To which of the field of two But does it belong? The detection bit value of the a-th bit position is defined. the bits above also at this way of setting -- anti- -- Power with the uncertainty of Translate for example, 1 bit of a bit value -- as the flag for a bit inversion judging Uncertainty is solvable by using etc.
[0524] in the case of n= 2 -- a figure -- it is shown in 44 -- as -- lambda<sup>(1)</sup>Is 0*Arg<sup>(1)</sup>It is Topi. It does not change but is lambda.<sup>(</sup>Is pi*Arg<sup>(</sup>Since it is To 2 pi, it is e.<sup>j</sup>the Multiplied by angle of deviation -- 180 degrees -- rotation -- To. and these total it asks for lambda, and on a complex plane, it is in the starting point and carries out an Arg sexagenary-cycle rectangular cross -- direct dividing a complex plane into two fields by making a line into a boundary line -- the bit value of one field -- " -- 1 consider it as "-- setting the field of another side to "0" -- change processing before lambda<sup>(1)</sup>lambda<sup>(</sup>It belongs to which of But and two fields. By whether it is To, the detection bit value of each bit position is defined.
[0525] Explain why the above detecting methods act well. At the time of embedding, it is each bit. To the complex embedding series for positions, it responds to a bit value and is +. It modulated by multiplying by 1 or 1. Therefore, complex correlation value of each bit position a of every lambda<sup>omega</sup>Is embedded rare Bit If A values differ, a phase The value from which only pi shifted will be taken. However, the same in a phase gap of pi They will be all if it sees and carries out. lambda<sup>(a)</sup>The amount of phase gaps of Is the input signal It will gather in the direction of delta theta. The same As a method of seeing, it is clear to perform the above change processings.
[0526] Length m of the complex embedding series for every bit is the length of all the complex embedding series! It is shorter. that is, since the complex embedding series for every bit has the low dispersion ratio, if the bit value for every bit is detected, it will be based on dispersion ratio m -- a profit is carried out and, for power, [ have ] to obtain, and a reason, tolerance becomes low. However, are above. Change which identifies a phase gap of pi in the same category is given, and they are all. lambda<sup>(a)</sup>Orientation After arranging, it totals. They are all the embedding series length by asking for lambda! It being alike and obtaining a corresponding profit It can do. Therefore, the boundary line which carries out an Arg sexagenary-cycle rectangular cross is used, and it is each Huh before change again.<sup>(a)</sup>This to evaluate rather than detecting for every bit by detecting a detection bit value -- a bit judging error -- few -- It is lost and higher tolerance can be realized. In the size of the above-mentioned absolute value to obtain It is even if it evaluates the reliability of digital-watermarking detection.
[0527] Feature > of the embodiment of Mr. 4
Next, the feature of this embodiment is explained.
[0528] According to the digital-watermarking detecting device of this embodiment, in detection of electronic Power, also when the synchronization of the signal for [ of electronic Power ] detection has shifted, detection of electronic Power can be performed. Namely, N -- Embedding series by which spectral diffusion was carried out in one-dimensional space, Direction of eye N dimension It uses that it is subject to common influence to the synchronous gap of Toward, and is N. -- In one-dimensional space By using the complex correlation value of the diffused series, it is unnecessary Electronic penetration of matching synchronization. Detection of Power can be performed.
[0529] For example, it is which can detect electronic Power, without using the method of special matching synchronization, also when the frame which starts detection in the direction of time has shifted in the case of the picture signal. This detects image Power ゝ electronic PenetrationゝTo re-photoed using the video camera etc., for example. Image power once changed into the case where it carries out, and analog data, such as videotape Electronic penetration When detecting To, in a difficult using state, a time synchronization is very effective.
[0530] Take a photograph with cameras displayed on a screen, TV, etc., such as an image power S video camera and a mobile phone. When carried out, since the reproductive frame rate and the frame rate of photography do not synchronize, a re-sampling with a subframe may produce them. This expresses with a subframe level (interval shorter than one frame) as a result the state where the synchronization has shifted. such a situation it is possible to also set and to measure the phase of a periodic signal in a time recovery -- the above -- Described -- detection of unnecessary electronic Power of matching synchronization is [ like ] possible.
[0531] While the efficient detection to the signal from which the synchronization shifted by the above-mentioned digital-watermarking detection is possible, To a matching synchronization signal since it is not necessary to add a special matching synchronization signal The quality in which degradation of a signal and degradation of the detection performance of electronic Power to depend drop off is high, and detection performance is high. Digital-watermarking detection is attained.
[0532] [Embodiment of The 5]
The amount measurement > of Synchronization displacement
Below, the digital-watermarking detecting device in the embodiment of The 5 is explained.
[0533] Electronic Penetration [ in / embodiment / this / in the embedding of electronic Power / the embodiment of The 1 of the present invention ] When carried out using embedding device 100, digital-watermarking detecting device smell the case where The and the signal with which the synchronization shifted in the direction of a time-axis (axis of the Nth dimension) are inputted -- said -- It is an example which detects the amount of displacement of a term and detects electronic Power.
[0534] In addition at this embodiment, it is, power in which the case where embedding of electronic Power is performed using electronic watermark embedding device 100 in the embodiment of The 1 is indicated for the example others -- operation Embedding of electronic Power is performed using the electronic Penetration embedding device in a form. Also detection of a It was case It can combine similarly and can apply. For example, The 3 The electronic watermark embedding device in an embodiment performs embedding, and it is a form of operation of The 3. N of the digital-watermarking detecting device in voice -- One-dimensional Fourier transform part 225 is combined, and it is Inspection. It may be made to perform appearance. When a suitable change is required for each procedure in those cases Although it is, explanation of this embodiment and they carry out electronic Penetration of those required change, and it buries them. If based on explanation of One and The at a lump, also until it will say a clear thing /!
[0535] Digital watermark detecting device >
The composition of the digital-watermarking detecting device in this embodiment is explained.
[0536] a figure -- 45 shows the example of composition of the digital-watermarking detecting device in the embodiment of The 5 of the present invention.
[0537] As for digital-watermarking detecting device 300 shown in the figure, it comprises time demodulation section 310, synchronous primary detecting element 320, detection information extraction part 330a, and pattern memory part 340, and embedding finishing signal 923 is ON. Power is carried out and detection information 914 is outputted.
[0538] About time demodulation section 310, it is the same as that of time demodulation section 210 in the embodiment of The 1. The time demodulation section of other embodiments may be used. For example, operation of The 2 Time demodulation section 210c of a form and 210 d may be used.
[0539] in addition -- a figure -- in 45 -- a figure -- which composition was indicated that information flows upwards from the bottom, and was noticed about and To in order to make an understanding of correspondence with 10 easy,
[0540] Detection processing of electronic Power by digital-watermarking detecting device 300 is carried out in the following procedures.
[0541] a figure -- 46 is a flow chart of operation of the digital-watermarking detecting device in the embodiment of The 5 of the present invention.
[0542] Step 1401 In time demodulation section 310, it restores to the direction of a time-axis, detection complex pattern 1501 is obtained, and it stores in pattern memory part 340. The contents of processing are operation of The 1. It is the same as that of time demodulation section 210 of digital-watermarking detecting device 200 of a form.
[0543] in addition in advance of the time recovery processing by time demodulation section 310, it may pretreat to embedding finishing signal 923 -- digital-watermarking detecting device 200 of the embodiment of The 1 -- said -- It needs. [0544] Step 1402 In synchronous primary detecting element 320, It is obtained by time demodulation section 310, and is an account of a pattern. Size of synchronous displacement in the direction of the time-axis (axis of the Nth dimension) currently beforehand added from detection complex pattern 1501 stored in part 340 to embedding finishing signal 923 It detects and outputs as an amount of synchronous displacement.
[0545] Mention the details of operation of synchronous primary detecting element 320 below.
[0546] Step 1403 In detection information extraction part 330a, It is obtained by time demodulation section 310, and the detection complex pattern stored in pattern memory part 340 is analyzed, obtaining by synchronous primary detecting element 320 , -- based on the amount 1502 of Was done synchronous displacement, it was embedded with electronic watermark embedding device 100 Electronic Penetration information is extracted and it outputs as detection information 914.
[0547] Mention the details of operation of detection information extraction part 330a below.
[0548] Digital watermark detecting device all the persons concerned term primary detecting element >
Next, the details of operation of the above-mentioned synchronous primary detecting element 320 are explained.
[0549] a figure -- 47 shows the example of composition of the synchronous primary detecting element in the embodiment of The 5 of the present invention.
[0550] Synchronous primary detecting elements 320 are complex detection series extraction part 321, complex correlation value calculating part 322, and an absolute value. It comprises calculation part 323, synchronous detection maximum judgment part 324, and phase calculation part 325, detection -- double -- base -- pattern 1501 is read from pattern memory part 340, and the amount 1502 of synchronous displacement is outputted
[0551] in addition -- a figure -- in 47 -- a figure -- which composition was indicated that information flows upwards from the bottom, and was noticed about and To in order to make an understanding of correspondence with 45 easy,
[0552] Synchronous detection processing by synchronous primary detecting element 320 is carried out in the following procedures.
[0553] a figure -- 48 is a flow chart of operation of the synchronous primary detecting element in the embodiment of The 5 of the present invention.
[0554] Step 1501 Detection complex putter inputted in complex detection series extraction part 321 Detection complex number series 1511 which put in order the complex number value acquired from A 1501 is constituted.
[0555] Operation of complex detection series extraction part 321 is the same as that of operation of detection series extraction part 221 in digital-watermarking detecting device 200 of the embodiment of The 4.
[0556] Step 1502 In complex correlation value calculating part 322, Complex [ which was constituted based on detection complex number series 1511 which comprised detection series extraction part 321, and the embedding series assumed ] The complex correlation with a number series is calculated and complex correlation value 1512 denoted by a complex number is calculated.
[0557] It is it about the complex correlation with a plurality of complex number series constituted based on a plurality of embedding series considered when the value which changes with kinds of embedding series was embedded. But calculation is done and corresponding complex correlation value 1512 is calculated.
[0558] Embedding series power in electronic watermark embedding device 100 For example, Embedding (Example 4) A plurality of symbol or a plurality of Bit as shown by series generation part 111 (Example 3) It comprises A and is in To go case, Embedding series corresponding to some of symbols or bits Based! It is ! even if it calculates the complex correlation with /and the Composed of(ed) complex number series. That is It is equivalent to uniting a synchronization using Chi, a plurality of symbols, or a part of bit.
[0559] Operation of complex correlation value calculating part 322 is the same as that of operation of complex correlation value calculating part 226 in digital-watermarking detecting device 200 of carrying of operation of The 4.
[0560] Step 1503 In absolute value calculation part 323, it was obtained by complex correlation value calculating part 322. Absolute value 1513 of complex correlation value 1512 is computed.
[0561] operation of absolute value calculation part 323 -- digital-watermarking detecting device 200 of the embodiment of The 4 -- It is the same operation as absolute value calculation part 227 to kick.
[0562] in step 1504 synchronous detection maximum judgment part 324 -- absolute value calculation part 323 -- profit Was done absolute value 1513 finds the thing used as the maximum, and is equivalent to absolute value 1513 used as the maximum -- double -- base -- correlation value 1512 is determined.
[0563] A constitution method of an embedding series [ in / in addition / electronic Penetration embedding device 100 ]
Instead of the maximum judging by synchronous detection maximum judgment part 324, it judges by other methods. It may be made like.
[0564] Mention the details of operation of synchronous detection maximum judgment part 324 below. Synchronous detection maximum About other methods used as substitution of judgment part 324, it is digital-watermarking detection of the embodiment of The 4. It is the same as that of the case of maximum judgment part 223 in device 200.
[0565] in step 1505 phase calculation part 325 -- synchronous detection maximum judgment part 324 -- determination -- computing the phase of a Was done complex correlation value and computing the amount 1502 of synchronous displacement based on this -- Detection information It outputs to news extraction part 330.
[0566] Mention the details of operation of phase calculation part 325 below. [0567] Digital-watermarking Embedding [ in / in addition / in the embedding of digital watermarking / the embodiment of The 3 ] When carried out using Place 100, N [ in / in advance of processing of complex detection series extraction part 321 / the embodiment of The 3 ] -- The same Fourier transform processing as one-dimensional Fourier transform part 225 It is needed.
[0568] Digital watermark detecting device all the persons concerned term primary detecting element all the persons concerned term detection maximum judgment part >
next, the details of operation of synchronous detection maximum judgment part 324 of the above-mentioned synchronous primary detecting element 320 -- just -- Explanation is carried out.
[0569] Digital watermark inspection [ in / in operation of synchronous detection maximum judgment part 324 / the embodiment of The 4 ] maximum judgment part 223 of appearance device 200 -- abbreviated -- same power It embeds as a result and acquires a series. It differs in that instead an absolute value calculates the complex correlation value used as the maximum.
[0570] Processing in synchronous detection maximum judgment part 324 is carried out in the following procedures.
[0571] Absolute value 1513 | rho obtained by 1 absolute-value calculation part 323<sup>(1)</sup>| | rho<sup>(2)</sup>| -- to a value -- the maximum -- Absolute value I p which becomes large (j) I is found.
[0572] I p<sup>(max)</sup>I = MAX (I p<sup>(1)</sup>| , | p<sup>(2)</sup>| - - - (98)
However, MAX O is an operation which returns the maximum.
[0573] 2) I p<sup>(max)</sup>It is a Noodle complex correlation value to the origin of I.<sup>max)</sup>It outputs to phase calculation part 325.
[0574] The greatest correlation value I p<sup>omega</sup>It judges whether I is over the predetermined threshold, and is predetermined. A threshold is exceeded. Electronic Power is embedded at and a case, and it judges with and a It is a force thing. It may be made like.
[0575] Digital watermark detecting device all the persons concerned term primary detecting element 1 phase calculation part >
Next, the details of operation of phase calculation part 325 of the above-mentioned synchronous primary detecting element 320 are explained.
[0576] Processing in phase calculation part 325 is carried out in the following procedures.
[0577] One Complex correlation value p acquired by synchronous detection maximum judgment part 324<sup>(max)</sup>It asks for Argument of △ theta. To.
[0578] delta theta =Arg[p<sup>(max)</sup>] (99)
However, Arg[] is an operation which searches for the angle of deviation of a complex number.
[0579] Since 2 △ 0 expresses the amount of gaps of a phase, calculate amount of synchronous displacement 1502At as follows, and it outputs it after this.
[0580] [74 number]
(1 0 0)
2 pi
However, tau is a cycle of a periodic signal.
[0581] Explain below the point that △ theta expresses the amount of gaps of a phase.
[0582] Complex correlation [ in / in a complex correlation value / digital-watermarking detecting device 200 of the embodiment of The 4 ] As explanation of One and The described to value calculation part 226, it is obtained like a following formula! /, To.
[0583] [75 number]
<img file="WO2007102403A1_D0049.tif" />
It follows,
Arg[p<sup>(max)</sup>] = A 0 (102)
It comes out, and it is and this shows the amount of gaps of the phase determined depending on the amount of synchronous displacement given to embedding finishing signal 923 in the direction of the axis of eye N dimension (for example, for method of time ).
[0584] In addition, it is electronic Penetration. Embedding series power in embedding device 100 For example, embedding system (Example 2) of sequence generation part 111, It is a bit by the difference in the positive/negative of an embedding series as shown in (Example 4). It is constituted and embedded and is in a and To case so that 0/1 of values may be expressed, synchronous gap the information to which the case where the signal with which half a wave of phase shifted is inputted, and all the bit values reversed at least The -- Fill -- Included -- rare -- ing -- a case -- distinction does not stick. That is, half a wave shifted from the amount At of synchronous displacement obtained above.
[0585] [76 number]<img file="WO2007102403A1_D0050.tif" />
That distinction which is the amount of synchronous displacement with which [ right ] does not stick.
[0586] the direction which in such a case embeds so that 1 ( to spread is 0) may certainly be taken by making the inner 1 bit of a bit value into the bit for a judgment, for example, constitutes the series, and serves as the value with the right bit value -- said -- It may be judged that it is the amount of term displacement. Numerals Y is carried out using an asymmetrical error correcting code. It is good also as a judgment being possible at things. Different digital watermarking and Trust from digital watermarking of the present invention It may be made to judge using an item. It judges as a value provisionally [ while ] as an amount of synchronous displacement, and is Bi by the above procedures in processing by detection information extraction part 330a. It may be made to amend Tut reversal. Others which drop off in what is limited to these examples Even if it amends by a method, also until right and a thing say /!
[0587] Diffuse the partial sequence of an embedding series as shown in (Example 4) of embedding series generation part 111 in the polarity of positive/negative for every business and The bit, and it is in a and To case, the last of the embodiment of The 4 -- a figure -- every bit position computed in the bit value detecting method explained with reference to 44 -- double -- Matter Correlation value<sup>(a)</sup>The sum of angle of deviation Arg of lambda -- the amount of synchronous displacement being referred to as delta theta -- each bit each time -- synchronization more certain and higher-precision than calculating the amount of synchronous displacement based on a complex correlation value Measurement of the amount of displacement is attained. Power which has the uncertainty of bit inversion also by this method By using 1 bit for example, of a bit value as a flag for a bit inversion judging, it is uncertainty. It is solvable.
[0588] Digital watermark detecting device Detection information extraction part >
Next, the details of operation of the above-mentioned detection information extraction part 330a are explained.
[0589] a figure -- 49 shows the example of composition of the detection information extraction part in the embodiment of The 5 of the present invention.
[0590] Detection information extraction part 330a, It is the same composition as detection information extraction part 220 of the embodiment of The 1, detection series extraction part 331, correlation value calculating part 332, maximum judgment part 333, and detection information -- re--- comprising formation part 334 -- detection information extraction part 220 of the embodiment of The 1 -- the amount 1 of synchronous displacement 502 is inputted and differs only from Point. [0591] in addition -- a figure -- in 49 -- a figure -- in order to make an understanding of correspondence with 45 easy -- the top from the bottom -- information which composition was indicated to flow and was noticed about and To,
[0592] As for the detection information extraction processing by detection information extraction part 330a, detection series extraction part 331 operates. Detection information extraction place [ in / except for details / detection information extraction part 220 of the embodiment of The 1 ] It is the same as that of Reason.
[0593] When embedding of digital watermarking is performed in addition using electronic watermark embedding device 100 in the embodiment of The 3, Processing of detection series extraction part 331 is preceded, and it is a fruit of The 3. N in the form of Application -- The same Fourier transform processing as one-dimensional Fourier transform part 225 is required. It becomes.
[0594] Digital watermark detecting device Detection information extraction part Detection series extraction part >
It is an opinion about the details of the operation of detection series extraction part 331 of detection information extraction part 330a to the following. Ming is carried out.
[0595] Processing by detection series extraction part 331 is carried out in the following procedures.
[0596] One From detection complex pattern 1501 to a size M X M chi - .. chi mu nu -- Complex [ one-dimensional ]
1 2 N- 1
Arrangement is constituted. Detection series extraction [ in / in the method of composition / the embodiment of above-mentioned The 1 ] It is the same as that of Step 801 in part 221.
[0597] Two The complex number sequence which took out and arranged every one complex number value sequentially from the complex arrangement obtained by above 1 is obtained. That is, complex arrangement alpha[rho and rho (rho*0),
1 2 --, Rho]
N- It is expressed 1 eta. When it is,
C =alpha [0, 0, - .., 0]
C =alpha [1, 0, - .., 0]
2
Embedding series 9 used at the time of the embedding in electronic watermark embedding device 100
It is L' =LZ2 when setting the length of 13 to L.
[0598] Three The amount 15 of synchronous displacement into which the phase of each element of the complex number sequence obtained by above 2 was inputted
You make it reverse-displaced based on 02. Namely, the amount 1502 of synchronous displacement
[0599] [77 number] 2 pi
Carrying out is,
[0600] [78 number]
c --c\e A'
-iotadeltatheta
C2 = *e<sup>J</sup>(1 04)
4) Take out and take every one complex number value sequentially from the complex number sequence obtained by above 3, and come out. The real part of the complex number value carried out and an imaginary part are seen as an independent real value, respectively, and are put in order. That is, detection series 1521 When expressed gamma and ...,
1 2 L
[0601] [79 number]
'<sup>3</sup>. 5)
'3 =<sub>2</sub>J
il =3[c<sub>2</sub>]
however
[0602] [80 number]
*,3
It is an operation which takes out the Each real part of Is a complex number, and an imaginary part.
[0603] Five It was obtained. It is Output to correlation value calculating part 332 considering gamma, gamma, ..., gamma as detection series 1521. To.
[0604] It is the order when exchange of an order of embedding series 913 or the element of a complex array is made in complex arrangement generation part 112 of electronic watermark embedding device 100. The point to return is the same as that of detection series extraction part 221 in the embodiment of The 1.
[0605] When embedding of digital watermarking is performed in addition using electronic watermark embedding device 100 in the embodiment of The 3, In above 2, it can set to the embodiment of The 3. Complex number of the element within the limits used by complex arrangement generation part 112 of the electronic Penetration embedding device A value is taken out and it is made to arrange.
[0606] Other examples > of composition of a Detection information extraction part
It is detection series extraction part 3, without using maximum judgment part 333 for detection information extraction part 330a. It constitutes only from 31, correlation value calculating part 332, and detection information reconstruction section 334, and they are the following processings. Detection information may be extracted by carrying out. such an example of composition -- a figure -- it is shown in 50.
[0607] a figure -- detection information extraction part 330b shown in 50 the complex correlation value greatest by synchronous primary detecting element 320 while reconstructing detection information with the acquired embedding series -- synchronous primary detecting element 320 -- profit Correlation value for evaluating the reliability of the detection information based on the amount 1502 of Was done synchronous displacement It recalculates anew. electronic Power can be detected at thereby more high speed It becomes obtaining.
[0608] One Processing of detection series extraction part 331 is the same as above-mentioned processing.
[0609] Two In correlation value calculating part 332, It is Synchronous detection in correlation value calculating part 222 in the embodiment of The 1 instead of differing and calculating a correlation value with all the embedding series considered. In appearance part 320, if the absolute value greatest by processing of synchronous detection maximum judgment part 324 is taken -- seal the embedding series corresponding to the complex correlation value by which Disconnect was carried out is memorized in a memory (not shown) -- a correlation value only with the embedding series of a correlation value calculating part 332 smell lever is calculated.
[0610] Judge whether reliable digital-watermarking detection was completed by whether this correlation value is larger than a predetermined threshold.
[0611] Three In detection information reconstruction section 334, the embedding series used by above 2 is used. If the point which reconstructs Detect information is removed, it is detection information reconstruction section 2 of the embodiment of The 1. It is the same as that of 24. [0612] Feature > of the embodiment of Mr. 5
according to digital-watermarking detecting device 300 of this embodiment, it sets to detection of electronic Power -- the time of the synchronization of the signal for [ of electronic Power ] detection having shifted -- the digital-watermarking signal itself It can use and can detect the amount of synchronous displacement. Namely, N -- It is Spatato in one-dimensional space. Embedding series by which Le diffusion was carried out, It is common influence to the synchronous gap by the direction of eye N dimension. It uses winning popularity and is N. -- The complex correlation value of the diffused series in one-dimensional space is used. By carrying out, detection of easy and electronic Power which can be matched the synchronization at high speed can be performed.
[0613] For example, it is which can detect electronic Power, without using the method of special matching synchronization, also when the frame which starts detection in the direction of time has shifted in the case of the picture signal. This detects image Power ゝ electronic PenetrationゝTo re-photoed using the video camera etc., for example. Image power once changed into the case where it carries out, and analog data, such as videotape Electronic penetration When detecting To, in a difficult using state, a time synchronization is very effective.
[0614] Take a photograph with cameras displayed on a screen, TV, etc., such as an image power S video camera and a mobile phone. When carried out, since the reproductive frame rate and the frame rate of photography do not synchronize, a re-sampling with a subframe may produce them. This expresses with a subframe level (interval shorter than one frame) as a result the state where the synchronization has shifted. such a situation it is possible to also set and to measure the phase of a periodic signal in a time recovery -- the above -- Described -- the amount of synchronous displacement is [ like ] detectable.
[0615] High-speed, since it cannot be based on the round robin technique of trying the amount of displacement one by one but calculation can detect the amount of synchronous displacement especially according to the digital-watermarking detecting device of this embodiment Efficient digital-watermarking detection is attained. Must which adds a special matching synchronization signal Degradation of a signal and Poor of the detection performance of electronic Power by a simultaneous arrival Matching signal since there is no important point The high digital-watermarking detection of the quality in which Y drops off with high detection performance is attained.
[0616] Like other examples of composition of a detection information extraction part mentioned above, if detection information extraction part 330b is constituted, it will be electronic Penetration at high speed. It becomes possible to detect.
[0617] [Embodiment of The 6]
At least To is phase abnormal conditions >.
It is an opinion about an electronic watermark embedding device [ in / to the following / the embodiment of The 6 of the present invention ]. Ming is carried out.
[0618] In electronic watermark embedding device 100 in the embodiment of The 1, this embodiment shows the example which performs abnormal-conditions processing in time modulation part 130 using delay of a periodic signal.
[0619] The composition of the electronic watermark embedding device in this embodiment is an embodiment of The 1. It is the same as that of electronic Penetration embedding device 100 which can be set, and only time modulation parts 130 differ. It is.
[0620] In addition, illustrate by this embodiment based on the embodiment of The 1! /and To are time modulation parts 1. The composition of other embodiments may be used in the composition of those other than 30. For example, in complex pattern generation part 110, it is an electronic watermark embedding device of the embodiment of The 3. Even if it uses complex pattern generation part 110b which can be set, it is right .
[0621] Digital watermark embedding device 1-hour modulation part >
Drawing 51 shows the example of composition of the time modulation part in the embodiment of The 6 of the present invention.
[0622] It comprises periodic signal generating part 131 and modulation part 136, and embedding complex pattern 921 are inputted, and time [ to be shown in the figure ] modulation part 130c outputs embedding pattern 922.
[0623] Generation processing of embedding pattern 922 by time modulation part 130c is carried out in the following procedures.
[0624] One A periodic signal is generated in periodic signal generating part 131. Example of a periodic signal [ in / in the periodic signal to generate / periodic signal generating part 131 of time modulation part 130a of a 1st embodiment ] It is the same.
[0625] in 2 modulation part 136, the periodic signal generated by above 1 is modulated as follows according to the complex number value of inputted embedding complex pattern 921 -- obtain Embedding Pata of N dimension - A 922.
[0626] Determine the amplitude of a periodic signal according to the absolute value of a complex number value.
[0627] According to the angle of deviation of a complex number value, delay a periodic signal, namely, change a phase.
[0628] Next, explain the example of the above-mentioned time abnormal conditions.
[0629] The abnormal conditions in modulation part 136, N -- According to a complex number value, the periodic signal generated by periodic signal generating part 131 is made into a subcarrier for every position of one-dimensional complex pattern 921, and it is QAM. It is carried out by changing into the pattern of N dimension in modulating (direct Intervals). [0630] However, the periodic signal which is a subcarrier is not necessarily a sine wave, as mentioned above.
[0631] When power is also constituted, make it only a real value change the phase of a periodic signal according to the real value, and all the values of embedding complex pattern 921 are the amplitude of a periodic signal. Be made to suppose that it is fixed.
[0632] It is specifically, for example, as follows, carried out.
[0633] Now and N -- A one-dimensional complex pattern It shall be expressed with rho (chi, X, - - -, X). This
1 2 N-1
The real part of P and an imaginary part shall be denoted by P and P the time of of.
[0634] [81 number]
rho (chiiota, chiiota, -, chi)<sub>nu</sub>^) -- = beta<sub>chi]</sub>chi<sub>2</sub>-, chi<sub>nu</sub>_,) e<sup>jaTixi</sup>(1 0 6)
It carries out. however, j is an imaginary unit -- omega is the angular velocity of the fundamental frequency of a periodic signal.
[0635] In periodic signal generating part 131, it is periodic signal f. (t) should be generated.
[0636] It is f about B and tau. It modulates by (t) and obtains pattern M of N dimension with a following formula.
[0637] [82 number]
Method of the time when baseband signal P drops off in the direction of time unlike general QAM abnormal conditions N which intersects perpendicularly with Toward -- It is changing in the one-dimensional direction (for example, the direction of space in the case of a picture signal). Be careful of To.
[0638] By such time abnormal conditions, the phase of embedding pattern 922 of N dimension is N. -- It will be spread so that it may change with positions on one-dimensional space, and it embeds at the time of digital-watermarking detection, and the size of the noise ingredient which originates in last signal 912 and appears becomes smaller.
[0639] Feature > of the embodiment of Mr. 6
the electronic watermark embedding device of this embodiment -- Digital watermarking of the embodiment of The 1 Because -- the example of composition from which time modulation part 130a in a lump device differs is shown -- The 1 -- real -- It has the feature in digital watermarking of the form of Application, and the same feature.
[0640] Especially in time modulation part 130c, it is N. -- N by which spectral diffusion was carried out to one-dimensional space -- One Embedding pattern of a dimension, the direction of eye N dimension intersect perpendicularly with it -- the phase of a periodic signal, and A it is modulating using an opposite value -- as opposed to the synchronous displacement given in the direction of eye N dimension -- N -- 1st order Have common influence in the original space! /and To! /-- it obtains and has the feature.
[0641] N which intersects perpendicularly with the direction of time in time modulation part 130c -- It is a one-dimensional direction (for example, the direction of space in the case of a picture signal), This from which it is made for the phase of embedding pattern 922 to differ It comes out. becoming less than the minimum picture signal quantization value in the case of a picture signal -- actual -- The embedding of digital watermarking should do. it can prevent a frame occurring -- Electronic penetration while being able to use the picture signal as a transmission way of Power effectively -- the size of the amplitude of digital watermarking -- coming If it is and changes targeting a frame, the tolerance over Attack can be increased when.
[0642] Embedding pattern N -- The phase is spread on one-dimensional space, And it depends, and becomes [ it embeds in as a result of correlation computation and the size of the noise ingredient which originates in a last signal and appears becomes smaller, and ] the reliable embedding of electronic Power, and detectable, and is the former. It becomes the embedding of electronic Power with less quality degradation, and detectable with comparable reliability.
[0643] [Embodiment of The 7]
Two or more rounds Time axis wave number zone embedding>
It is an opinion about an electronic watermark embedding device [ in / to the following / the embodiment of The 7 of the present invention ]. Ming is carried out.
[0644] This embodiment is Based to a periodic signal which is different in the digital-watermarking embedding of electronic watermark embedding device 100 in the embodiment of The 1! They are line and an example which embeds using long and a spectral diffusion series more, and embeds information at the time of /and Multiple said.
[0645] a figure -- an electronic watermark embedding device [ in / in 52 / the embodiment of The 7 of the present invention ], and electron The example of composition of a watermark detecting device is shown.
[0646] <Electronic watermark embedding device>
electronic Penetration shown in Drawing 52 Embedding device 500 is [ complex pattern generation part 510 and time ] strange. comprising tone part 520 and embedding pattern superposed part 530 -- embedding information 3111 -- burying Signal 3112 before a lump is inputted and embedding finishing signal 3113 is outputted.
[0647] In the procedure of the following [ processing / of digital watermarking by electronic watermark embedding device 500 / embedding ] It carries out.
[0648] a figure -- 53 is a flow chart of operation of the electronic Penetration embedding device in the embodiment of The 7 of the present invention.
[0649] Step 1601 It is Based to embedding information 3111 inputted in complex pattern generation part 510! /, The, and a plurality of embedding complex patterns 3121 are generated.
[0650] N which, as for each embedding complex pattern 3121, complex number power also comprises -- It is a one-dimensional pattern and expresses the contents of embedding information.
[0651] Mention the details of operation of complex pattern generation part 510 below.
[0652] Step 1602 In time modulation part 520, it embeds based on each embedding complex pattern 3121 generated in each complex pattern generation part 510, and generates pattern 3122.
[0653] Operation of time modulation part 520 is the same as that of time modulation part 130 in the embodiment of The 1. However, as for the periodic signal generated in each time modulation part 520, each intersects perpendicularly mutually. Suppose that it is a periodic function to carry out. For example, it is Buddy with the periodic function with which fundamental frequency differs, respectively. The is also good.
[0654] The time modulation part shown by embodiments other than the embodiment of The 1 may be used as time modulation part 520. For example, the embodiment of The 2 or the embodiment of The 6 Even if it uses a time modulation part, it is style trap .
[0655] When time modulation part 130b of the embodiment of The 2 is used as time modulation part 520 It is also possible to process the time abnormal conditions of a plurality of frequency by Fourier transform once. It is.
[0656] in step 1603 embedding pattern superposed part 530, it superimposes on signal 3112 before embedding into which each embedding pattern 3122 generated by each time modulation part 520 was inputted -- output embedding finishing signal 3113.
[0657] Mention the details of operation of embedding pattern superposed part 530 below.
[0658] Electronic power Embedding device-complex pattern generation part > Below, the details of operation of the above-mentioned complex pattern generation part 510 are explained.
[0659] a figure -- the example of composition of a complex pattern generation part [ in / in 54 / the embodiment of The 7 of the present invention ] -- To indicate.
[0660] It comprises embedding series generation part 511 and a plurality of complex arrangement generation parts 512, embedding information 3111 is inputted, and complex pattern generation part 510 is about embedding complex pattern 3121. It outputs.
[0661] In the procedure of the following [ processing / by complex pattern generation part 510 / embedding complex pattern generation ] It carries out.
[0662] a figure -- 55 is a flow chart of operation of the complex pattern generation part in the embodiment of The 7 of the present invention.
[0663] In step 1701 embedding series generation part 511, generate the numerical sequence showing embedding information based on inputted embedding information 3111, divide this, and plurality buries. Lump series 3211 is generated.
[0664] Mention the details of operation of embedding series generation part 511 below.
[0665] Step 1702 It is N about each embedding series 3213 generated by embedding series generation part 511 in complex arrangement generation part 512. - To the real part and imaginary part of an element on one-dimensional complex arrangement It assigns and generates embedding complex pattern 3121.
[0666] Complex arrangement generation [ in / in operation of each complex arrangement generation part 512 / the embodiment of The 1 ] It is the same as that of operation of part 112.
[0667] Complex pattern generation part [ in / for complex pattern generation part 510 / the embodiment of The 3 ] 1 It may be made to constitute based on 10b. namely, -- being obtained by complex arrangement generation part 512 N [ in / further / for a It was pattern / the embodiment of The 3 ] -- one-dimensional inverse Fourier transform part 113 -- said -- him -- it is good also as complex pattern 3121 which embeds the result Fourier-transformed by processing.
[0668] Electronic power Embedding device-complex pattern generation part Embedding series generation part >
Embedding series generation part [ in / at embedding series generation part 511 / the embodiment of The 1 ] 1 They are a plurality of portions about the embedding series after embedding with the same procedure as 11 and generating a series. It divides. For example, it is series w= by the same procedure as embedding series generation part 111. When {w, w, - .., and w} are generated, it is each embedding series 3213w.<sup>[1]</sup>, w<sup>C2]</sup>- .., w<sup>[n]</sup>To, [0669] [83 number]
w[<sup>2</sup>l = {<sub>Wl+1</sub>, w<sub>L+2</sub>,---,w<sub>2L</sub>) (1 08)
W<img file="WO2007102403A1_D0051.tif" />
It carries out.
[0670] However, n is a total of division.
[0671] In addition, embed here and it is head power about a series. Although the predetermined example divided every [ several ] was shown in order
As long as it divides with the split method decided beforehand, it may divide how. For example,
[0672] [84 number]
={w],w<sub>n+</sub>i<sub>!</sub>--,w<sub>a</sub>_i)<sub>n+1</sub>)
W[<sup>2</sup>] = {w<sub>2</sub>'W*+2 -■■-■ W(I-l) n+2 } (1 09)
It is good as for a method of of.
[0673] Digital watermark embedding device Embedding pattern superposed part >
Below, the details of operation of embedding pattern superposed part 530 are explained.
[0674] embedding pattern superposed part [ in / in operation of embedding pattern superposed part 530 / the embodiment of The 1 ] 140 -- abbreviated -- same power Only the following points differ.
[0675] With embedding pattern superposed part 530, It superimposes by adding each embedding pattern 3122 of N dimension generated by each time modulation part 520 to N dimension signal inputted as signal 3112 before embedding, N dimension signal of the superimposed result is embedded, and it is considered as ending signal 3113. Output is carried out. At this time, a plurality of embedding patterns 3122 are added altogether, and are superimposed. It is even if it emphasizes and superimposes each embedding pattern 3122 by different intensity as and embedding intensity. It does not matter. for example, the frequency band of each embedding pattern 3122 which performs embedding is received making it change embedding intensity, respectively, when the Deterioration characteristics differ -- each Fill -- detection of a lump pattern may be made to be performed in the same accuracy.
[0676] Digital watermark detecting device >
Digital-watermarking detecting devices 600 in this embodiment are a plurality of synchronous primary detecting elements 620 and Inspection. It comprises appearance information extraction part 630, embedding finishing signal 3113 is inputted, and it is detection information 3. 114 is outputted.
[0677] Digital-watermarking detection processing by digital-watermarking detecting device 600 is carried out in the following procedures.
[0678] a figure -- 56 is a flow chart of operation of the digital-watermarking detecting device in the embodiment of The 7 of the present invention.
[0679] Step 1801 In each time demodulation section 610, it restores to the direction of a time-axis, and obtains detection complex pattern 3161. The contents of the processing in each time demodulation section 610 are the embodiments of The 1. The same power as time demodulation section 210 in digital-watermarking detecting device 200 Each time recovery Periodic function used by time modulation part 520 of electronic watermark embedding device 500 every part 610 It uses, respectively.
[0680] In addition in advance of the time recovery processing by time demodulation section 610, it is with digital-watermarking detecting device 200 of the embodiment of The 1 that it may pretreat to embedding finishing signal 3113. It is the same.
[0681] The time demodulation section shown by embodiments other than the embodiment of The 1 may be used as time demodulation section 610. For example, time demodulation section 210c in the embodiment of The 2 You may use.
[0682] It is also possible to carry out time recovery processing of a plurality of frequency by Fourier transform once as time demodulation section 610, in using time demodulation section 210c of the embodiment of The 2, and it is Ah. To.
[0683] Step 1802 In each synchronous primary detecting element 620, From each detection complex pattern 3161 obtained by time demodulation section 610, it is beforehand added to embedding finishing signal 3113, respectively. The size of synchronous displacement is detected in the direction of the time-axis (axis of the Nth dimension) which was, and it is the amount 3 of synchronous displacement. It outputs as 162. [0684] Digital-watermarking detecting device [ in / in operation of synchronous primary detecting element 620 / the embodiment of The 5 ] 300 It is the same as that of synchronous primary detecting element 320.
[0685] in step 1803 detection information extraction part 630, it was obtained by time demodulation section 610 -- each -- said -- Electronic penetration embedded with electronic watermark embedding device 500 based on the amount 3162 of term displacement It carries out, information is extracted and it outputs as detection information 3114.
[0686] Mention the details of operation of detection information extraction part 630 below.
[0687] Digital watermark detecting device Detection information extraction part >
Below, operation of a detection information extraction part is explained in detail.
[0688] a figure -- 57 shows the composition of the detection information extraction part in the embodiment of The 7 of the present invention.
[0689] Detection information extraction part 630 shown in the figure is to detection information extraction part 330 of the embodiment of The 5. Each detection complex pattern 316 into which it is similar composition and detection series extraction part 631 was inputted
Easy is carried out every amount 3162 of synchronous displacement, and differs from Point.
[0690] in addition -- a figure -- in 57 -- a figure -- which composition was indicated that information flows upwards from the bottom, and was noticed about and To in order to make an understanding of correspondence with 54 easy,
[0691] In the detection information extraction processing by detection information extraction part 630, a plurality of detection series extraction parts 631 are.
It is based on detection complex pattern 3161 inputted, respectively and the amount 3162 of synchronous displacement, and is Inspection. The 5 carries out except for the point of extracting appearance series 3313, and the details of operation of correlation value calculating part 632. It is the same as that of the detection information extraction processing in detection information extraction part 330 of a form.
[0692] Like [ of other examples of composition of detection information extraction part 330 described by the embodiment of The 5 ], not using maximum judgment part 633, the greatest complex correlation value in synchronous primary detecting element 320 was acquired -- burying While reconstructing detection information with a lump series Synchronization obtained by synchronous primary detecting element 320 Based on the amount of displacement, the correlation value for evaluating the reliability of the detection information concerned is calculated anew. It may rebe made to carry out.
[0693] Digital watermark detecting device Detection information extraction part Correlation value calculating part >
Operation of correlation value calculating part 632 is explained in detail below.
[0694] Processing in correlation value calculating part 632 is carried out in the following procedures.
[0695] One Each detection series 3313i" acquired by detection series extraction part 631<sup>[1]</sup>gamma<sup>epsilon2]</sup>-- Integration It carries out and is a series. gamma is obtained. namely, [0696] [85 number]
<img file="WO2007102403A1_D0052.tif" />[k] indicated on the shoulder of gamma here, and k-th detection series extraction part 631 -- k-th Inspection it expresses that it is the detection series acquired from appearance complex pattern 3161 -- the detection into which n is inputted -- double -- base -- it is a total of pattern 3161. example which connects each detection series simply here it can set to embedding series generation part 511 of shown power electronic watermark embedding device 500 -- burying corresponding to the split method of a lump series -- If, It is a value at the order beforehand decided from each detection series. It takes out and a series is combined. For example, it joins together as follows. The is not cared about, either.
[0697] [86 number]<img file="WO2007102403A1_D0053.tif" />
2) Series acquired by above 1 Origin is asked for correlation value 3314 for gamma by the same processing as correlation value calculating part 332 of digital-watermarking detecting device 300 of the embodiment of The 5. however, correlation meter electronic watermark embedding device 500 buries the embedding series which is the target of Calculation and which is considered -- Included Series [ before seeing being generated in series generation part 511 and divided ] w= {w, w, - .., and w}
1 Note that it is 2 nL.
[0698] Improvement > in accuracy by integration of the amount of Synchronization displacement
The amount 31 of synchronous displacement obtained every detection complex pattern 3161 in synchronous primary detecting element 620 Based on 62, the higher-precision amount of synchronous displacement is calculated with the following procedures. It is good.
[0699] One Each amount 3162 of synchronous displacement,
[0700] [number 87] △ ■ r], delta lambda, *<sub>2</sub>--
2pi 2pi
It carries out. Here, tau, tau, and -- are the cycles of the periodic signal of each time demodulation section 610.
1 2
[0701] Let the greatest thing be one selection At from 2 At, At, and --. Or a cycle is the maximum.
1 2 max
At corresponding to a Period signal may also be chosen.
[0702] Three The equation of At, T, and At Power Next is considered to each i, and it asks for n.
max
[0703] [88 number]
delta*, 10", =At<sub>max</sub>(1 1 2)
4) It is each each about a Kinki integral value most to n. It is considered as eta'.
[0704] Ask for following At' using 5 eta'.
[0705] [89 number]
At<sub>t</sub>'=alpha*,<img file="WO2007102403A1_D0054.tif" />1 3)
this is united with the position of At for every periodic signal -- position of eta' cycle eye nu dimension max
It means calculating the amount of displacement in the direction of an axis of eyes as At'.
[0706] a figure -- the situation of the above-mentioned calculation at the time of using two periodic signals for 58 is shown. In the figure, it is △ theta = pi/to periodic signal 1 of cycle 6 to △ theta =5pi/3, and periodic signal 2 of cycle 4.
1 2
2 is obtained and this is expressed with a circle [ white ]. It is set to At =5 and At =1 at this time. Again
1 2
The position of At+T and At+2T is denoted by the black dot. At is chosen as At and it describes above.
2 2 2 n' = 1 is obtained by Procedure 3 of 2 max 1, and 4. About that of the black dot of the left-hand side among this and two black dots
2
The thing which are two periodic signals and which the point of △ 0 and △ 0 overlaps, respectively in a phase is denoted by Place. Result
2
delta*' =delta*+eta' tau =5 are obtained.
2 2 2 2 [0707] In addition about i corresponding to the periodic signal selected as At, it is set to n' = l.
max 1
[0708] Six The average value of At' obtained as mentioned above is calculated, and it is considered as the amount At of displacement as the whole.
[0709] [90 number]
delta* ; (1 1 4)
k =iota
However, k is the number of a periodic signal.
[0710] a figure -- although the example of 58 showed the example in case there is no error in detection of each amount of synchronous displacement, n may not become an integer when detection of each amount of synchronous displacement has an error. Above An error is taken into consideration by asking for n' as an integer in Procedure 4, and finally calculating the average value of At'. The maximum likelihood value of the amount of displacement in the axis of eye N dimension carried out will be calculated.
[0711] When calculating the average value of At', it may be made for the amount of displacement from which the position shifted to Brightness to be disregarded for example. Thereby for example, an attack is delivered to the frequency corresponding to a specific periodic signal, and detection of the amount of synchronous displacement obtained from detection complex pattern 3161 goes wrong. The case where carried out The and it waits can be removed and considered.
[0712] Since it is the value which was able to calculate the amount of displacement of the direction of eye actual N dimension with sufficient accuracy, 7 At is the amount of synchronous displacement of each periodic signal after this.
[0713] [91 number]<img file="WO2007102403A1_D0055.tif" />
Eye Asks, in addition To,
[92 number] 81 (in the case of lambda**";gamma) -- delta/_";
(1 1 5)
<sup>1</sup>(<<; - 1) '~alpha* - tau (alpha* < n T),<sub>t</sub>In the of case
This which asks for the amount of displacement of the direction of Based and eye a The N dimension the amount 3162 of synchronous displacement obtained every detection complex pattern 3161 as mentioned above with sufficient accuracy, and recalculates the amount of synchronous displacement using it When It is Inspection of a possible next door, a result, and digital watermarking to compute the higher-precision amount of synchronous displacement. Appearance accuracy can be raised.
[0715] Feature > of the embodiment of Mr. 7
electronic Penetration of this embodiment according to an embedding device and a digital-watermarking detecting device -- double -- Carry out electronic With power of the embedding information that information length is longer, by using the frequency band of a number. It can embed.
[0716] moreover -- each -- calculating a correlation value for the detection series of the result obtained by matching synchronization being performed separately as a whole finally, and evaluating the reliability of detection, Is it Electronic penetration separately? To is embedded, and reliability of the whole detection result is more correctly clarified rather than detecting. Things are made.
[0717] More reliable digital watermarking which can lengthen spectral diffusion series length more Embedding can be performed. This is explained below.
[0718] The above-mentioned literature "Kana Yamamoto, Takao Nakamura, Yoichi Takashima, Jun Katayama, According to Ryo Kitahara and * Taketaka "1 about detection quality assessment of frame superposition type video electronic Power consideration" information-science-and-technology forum F IT2005, J-029, and 2005" Place of electronic Power using spectral diffusion and correlation computation The detection evaluation value showing the reliability of detection in the meaning of the false positivity of Synthesis and electronic Power is a spectrum. In proportion to the square root of the series length of diffusion, it becomes large. On the other hand, it is plurality like this embodiment. When it superimposes on a frequency band and digital watermarking is embedded, in order to suppress the whole signal degradation It is necessary to lessen energy of the watermark signal for every frequency band. namely, the Into (for example, Into to which the value of PSNR is not changed) to which the grade of signal degradation is not changed -- frequency of n pieces the case where it multiplexes to a zone -- a frequency band with independent energy of each watermark signal -- Fill -- To include -- a case -- it compares, and is set to lZn and amplitude serves as 1Zn. This is each embedding. U [ serve as / intensity / 1Z n ],
[0719] The above-mentioned literature "Kana Yamamoto, Takao Nakamura, Yoichi Takashima, Jun Katayama, When embedding intensity serves as lZ n according to Ryo Kitahara and * Taketaka "1 about detection quality assessment of frame superposition type video electronic Power consideration" information-science-and-technology forum F IT2005, J-029, and 2005", For a watermark, it is not based on A , but becomes fixed. If the original picture ingredient used as A is small enough, the detection evaluation value of digital watermarking is embedding intensity. (epsilon[when [ The figure of the literature concerned ] the limit to which alpha becomes large, i.e., embedding intensity , becomes large enough compared with an original picture ingredient by 1)<sub>rho</sub>] is a watermark moreover it carries out asymptotic to 1. If the original picture ingredient which takes and serves as a noise is large enough, 1/of detection evaluation values of digital watermarking will be set to eta (epsilon[when [ The figure of the literature concerned ] the limit to which alpha becomes small, i.e., embedding intensity, becomes small enough compared with original picture ingredient by 1).<sub>rho</sub>] approaches the straight line which passes along the starting point.
[0720] When multiplexing to eta frequency bands as a result, compare with the case where it embeds at an independent frequency band,
If the original picture ingredient used as a noise is small enough for 1 watermark, it is detection evaluation of digital watermarking. A value becomes large in proportion to the square root of the series length of spectral diffusion,
Even when the original picture ingredient used as a noise is large enough for 1 watermark, is it Electronic penetration even when it is the worst? The detection evaluation value of To does not change.
It becomes things, and as the whole, a detection evaluation value becomes large and detection of it with reliability high as a result is attained.
[0721] It is based on the amount 3162 of synchronous displacement obtained every detection complex pattern 3161. It is business about the method of calculating the amount of displacement of the direction of nu dimension eye with sufficient accuracy, and recalculating the amount of synchronous displacement using it. If it is, It becomes possible to compute the higher-precision amount of synchronous displacement, and is detection of electronic Power. If accuracy can be raised and it is conversely comparable detecting accuracy, it is the embedding of electronic Power. It can weaken and can realize digital watermarking with little quality degradation.
[0722] As time modulation part 520 and time demodulation section 610, a plurality of cycles when using time modulation part 130b and time demodulation section 210c of an embodiment of The 2 of the present invention, and 210 d processing of the time abnormal conditions of a number and a time recovery can be once carried out by Fourier transform -- more -- high -- Fast processing is attained.
[0723] [Embodiment of The 8] Time multiplex embedding>
An electronic watermark embedding device and an electron A watermark detecting device is explained. [ in / to the following / the embodiment of The 8 of the present invention ]
[0724] An electronic watermark embedding device and an electron A watermark detecting device is used, [ in / in this embodiment / the embodiment of The 5 ] a plurality of Burial which follow a synchronous pattern and it to the signal before an input Because -- the embedding pattern based on lump information is embedded, and the amount of displacement of a synchronization is detected, and abundant It is an example which embeds embedding information efficiently and detects it.
[0725] <Electronic watermark embedding device>
Electricity [ in / in the electronic Penetration embedding device in this embodiment / the embodiment of The 1 ] Child permeability is carried out, and it has the same composition as embedding device 100, and is complex pattern generation part 110. A part of operations differ.
[0726] Embedding processing of digital watermarking by electronic watermark embedding device 100 in this embodiment is performed by the following procedures.
[0727] a figure -- operation of an electronic Penetration embedding device [ in / in 59 / the embodiment of The 8 of the present invention ] It is a flow chart.
[0728] Step 1901 Embedding information 91 inputted in complex pattern generation part 110
It embeds based on 1 and generates complex pattern 921. this time -- circumference of time modulation part 130 being based on embedding information for every cycle of the periodic signal which term signal generating part 131 generates -- Fill -- it generates so that lump complex pattern 921 may change.
[0729] Mention the details of operation of complex pattern generation part 110 below.
[0730] Step 1902 In time modulation part 130, it is generated in complex pattern generation part 110. It is stored in the 1st storage part 150, and they are Based and a Embed pattern to and Complex pattern 921.
922 is generated and it stores in storage part 160 of The 2.
[0731] Operation of time modulation part 130 removes the point that embedding pattern 922 changes for every cycle, according to embedding complex pattern 921 which complex pattern generation part 110 generated, and is The 1. It is the same as that of operation of time modulation part 130 in an embodiment.
[0732] In addition about time modulation part 130, the time modulation part of other embodiments may be used. For example, time modulation part 130b of the embodiment of The 2 may be used, and it is The 6. Even if it uses time modulation part 130c of an embodiment, it is style trap .
[0733] In step 1903 embedding pattern superposed part 140, It generates by time modulation part 130. It was carried out, embedding pattern 922 stored in storage part 160 of The 2 was inputted, and it buries. It superimposes on signal 912 before a lump, and outputs embedding finishing signal 923.
[0734] Operation of embedding pattern superposed part 140 is the same as that of the embodiment of The 1.
[0735] Digital watermark embedding device-complex pattern generation part >
Below, the details of operation of complex pattern generation part 110c are explained.
[0736] a figure -- the example of composition of a complex pattern generation part [ in / in 60 / the embodiment of The 8 of the present invention ] -- To indicate.
[0737] It comprises embedding series generation part 117, complex arrangement generation part 116, embedding information dividing part 114, and synchronous series generation part 115, embedding information 911 is inputted, and complex pattern generation part 110c outputs embedding complex pattern 921.
[0738] Generation processing of the embedding complex pattern by complex pattern generation part 110c is carried out in the following procedures.
[0739] a figure -- 61 is a flow chart of operation of the complex pattern generation part in the embodiment of The 8 of the present invention.
[0740] Step 2001 In synchronous series generation part 115, synchronous series 917 which is a sequence of the numerical value for matching synchronization decided beforehand is generated.
[0741] Mention the details of operation of synchronous series generation part 115 below.
[0742] Divide inputted embedding information 911 into a plurality of partial embedding information 916 in step 2002 embedding information dividing part 114. As for the method of division, anything is a style. Do not know. As [ divide / for example, / in each order / K bits of last power of embedding information 911 ] It may be.
[0743] In step 2003 embedding series generation part 117, it is a numerical sequence which expresses Based , The, and embedding information to partial embedding information 916 acquired by embedding information dividing part 114. Embedding series 913 is generated.
[0744] Operation of embedding series generation part 117 is each of a plurality of partial embedding information 916. The point which receives, embeds and generates series 913 is removed, and it is an embedding series of the embodiment of The 1. It is the same as that of generation part 111.
[0745] Step 2004 In complex arrangement generation part 116, synchronous series 917 generated by synchronous series generation part 115, and each embedding series 913 generated by embedding series generation part 117 -- respectively -- N -- assigning the real part and imaginary part of an element on one-dimensional complex arrangement -- a plurality of Fill -- lump complex pattern 921 is generated.
[0746] One and The mention below in the details of operation of complex arrangement generation part 116.
[0747] Digital watermark embedding device-complex pattern generation part all the persons concerned term series generation part >
In synchronous series generation part 115, the following processings generate synchronous series 115.
[0748] Synchronous series 115 is a sequence of a value used for matching synchronization in a digital-watermarking detecting device, and generate it using a pseudorandom-numbers sequence so that it may not overlap with other embedding series. Namely, Pseudo It is alike and is random number sequence SPN=. {SPN, SPN, - .., and SPN} (L is the length of a series) It is a synchronous series when carrying out.
1 Two It Carries Out.
s = {s, S, --, S}
1 Two It Carries Out.
s = SPN= {SPN, SPN, --, SPN} (116)
1 Two It Carries Out.
of -- it may determine like.
[0749] Electronic power and embed and it is device-complex pattern generation part-complex arrangement generation part >.
Operation of complex arrangement generation part 116 is an electronic watermark embedding device of the embodiment of The 1. Although it is similar with complex arrangement generation part 112 which can be set, Synchronous series 917 and a plurality of embedding The complex arrangement corresponding to each of series 913 is constituted, It is a cycle of time modulation part 130 about it. Embedding complex pattern 921 is for every cycle of the periodic signal which signal generating part 131 generates. It differs in that it generates so that it may be exchanged one by one, and they are and To.
[0750] In complex arrangement generation part 116, embed by the following processings and generate complex pattern 921.
[0751] One It is Based on to synchronous series 917 generated by synchronous series generation part 115 by the same procedure as complex arrangement generation part 112 in electronic watermark embedding device 100 of the embodiment of The 1. Embedding complex pattern SP is generated.
[0752] Two The same procedure as complex arrangement generation part 112 in electronic watermark embedding device 100 of the embodiment of The 1, Each embedding series 91 generated by embedding series generation part 117 Embedding complex patterns A and A and - .. based on 3, and alpha are generated. However, k is embedded. The number, i.e., the embedding information dividing part, of embedding series 913 generated by series generation part 117
It is the number of division of the information on 114.
[0753] Three For every cycle of the periodic signal which periodic signal generating part 131 of time modulation part 130 generates, it embeds in following order, and repeats and outputs a complex pattern.
[0754] SP, SP, A, A, - .., A, SP, SP, A, A, - .., and A --
1 2 k 1 2 k
(117)
-- of an end means that the whole is repeated similarly.
[0755] Although embedding complex pattern SP based on synchronous series 917 is repeated twice and outputted here, you may make it repeat this 3 times or more. in that case, electronic Power is detected supposing having repeated two or more times in a digital-watermarking detecting device -- as -- until it says carrying out --
[0756] Based and a Embed pattern are generated by embedding complex pattern 921 generated in this way, and embed embedding information by time sharing by the signal before embedding being overlapped. It becomes things.
[0757] a figure -- the example which continues and embeds a plurality of information by time sharing 62 is shown. As shown in the figure, the embedding pattern generated from each embedding complex pattern is connected, and it embeds, and is a last signal. It is superimposed.
[0758] Digital watermark detecting device >
Drawing 63 is Show about the example of composition of the digital-watermarking detecting device in the embodiment of The 8 of the present invention. To.
[0759] Digital-watermarking detecting device 700 shown in the figure, At the time of embedding finishing signal dividing part 710 and a synchronization Between demodulation section 720, synchronous primary detecting element 730, synchronized signal dividing part 740, time demodulation section 750, detection It comprises information extraction part 760 and pattern memory part 770, Embedding finishing signal 923 inputs. It is carried out and detection information 3812 is outputted.
[0760] in addition -- a figure -- in 63 -- a figure -- which composition was indicated that information flows upwards from the bottom, and was noticed about and To in order to make an understanding of correspondence with 10 easy,
[0761] Detection processing of electronic Power by digital-watermarking detecting device 700 is a flow chart of operation of a digital-watermarking detecting device [ in / in [0762] figure 64 carried out in the following procedures / the embodiment of The 8 of the present invention ].
[0763] Finishing of the embedding of the part of the same length as the cycle of the periodic signal in time modulation part 130 of electronic watermark embedding device 100 in step 2101 embedding finishing signal dividing part 710 Signal 923 is inputted and partial embedding finishing signal 3816 is acquired.
[0764] Step 2102 In synchronous time demodulation section 720, Digital-watermarking detecting device [ in / to partial embedding finishing signal 3816 divided at Step 2101 / the embodiment of The 1 ] With the same procedure as time demodulation section 210, it gets over to a complex pattern and makes this into Synchronous complex butter - A 3813.
[0765] In addition about synchronous time demodulation section 720, it is good also as the same operation as the time demodulation section in the other embodiments of the present invention. For example, you may make it be the same as that of operation of time demodulation section 210c of the embodiment of The 2.
[0766] Step 2103 In synchronous primary detecting element 730, Synchronous detection of digital-watermarking detecting device [ in / to synchronous complex pattern 3813 obtained at Step 2102 / the embodiment of The 5 ] 300 The amount 3814 of synchronous displacement is calculated with the same procedure as appearance part 320.
[0767] The amount of synchronous displacement can be calculated in synchronous primary detecting element 730, and it is Step 21 in a It is a force case.
Processing is repeated to a back embedding finishing signal by return and 1 cycle to 01.
[0768] Mention the details of operation of synchronous primary detecting element 730 below.
[0769] Step 2104 In synchronized signal dividing part 740, From the position to which the part of the amount 3814 of synchronous displacement obtained at Step 2103 and embedding finishing signal 923 were shifted, carrying out in Electronic penetration and burying -- Included See, embed and end with the same length as the cycle of the periodic signal in time modulation part 130 of device 100. dividing signal 923 -- embedding information dividing part 114 of electronic watermark embedding device 100 -- part A number of embedding information of synchronized partial signals 3817 by which the rate was carried out are acquired.
[0770] Mention the details of operation of synchronized signal dividing part 740 below.
[0771] Step 2105 In time demodulation section 750, As opposed to each synchronized partial signal 3817 divided at Step 2104, At the time of digital-watermarking detecting device 200 in the embodiment of The 1 It gets over to a complex pattern with the same procedure as between demodulation section 210, and is a detection complex putter about this. It is considered as A 3815 and stores in pattern memory part 770. [0772] Time recovery in the embodiment [ demodulation section / 750 / time ] of the others of the present invention in addition It is good also as the same operation as a tone part. For example, Motion of time demodulation section 210c of the embodiment of The 2 You may make it be the same as that of Work.
[0773] Step 2106 In detection information extraction part 760, Digital-watermarking detecting device [ in / to each detection complex pattern 3815 obtained at Step 2105 / the embodiment of The 1 ] 200 After acquiring each detection information with the same procedure as detection information extraction part 220, Each Inspection Appearance information is connected and the whole detection information 3812 is outputted.
[0774] Mention the details of operation of detection information extraction part 760 below.
[0775] Digital watermark detecting device all the persons concerned term primary detecting element >
Below, the details of operation of synchronous primary detecting element 730 are explained.
[0776] The power as synchronous primary detecting element 320 of digital-watermarking detecting device 300 in the embodiment of The 5 in which the operation in synchronous primary detecting element 730 is the same The following points differ.
[0777] In complex correlation value calculating part 322 in the embodiment of The 5, It is this operation while the complex correlation with the complex number series constituted based on the embedding series assumed was calculated. With complex correlation value calculating part 322 of a form, Synchronous series student of electronic Penetration embedding device 110c Complex phase with the complex number series constituted based on synchronous series 917 generated in Formation part 115 Seki is calculated.
[0778] In synchronous detection maximum judgment part 324, when the value of absolute value 1513 is not over the predetermined threshold, a synchronous pattern can be detected, and judge it as a It is a force thing, and it is the amount 150 of synchronous displacement. 2 is not outputted.
[0779] Perform repetition processing until the amount 3814 of synchronous displacement is calculated from the embedding finishing signal started at a time one by one one cycle by embedding finishing signal dividing part 710 as mentioned above. A synchronous series will embed to Find power and will scan a settled signal one by one.
[0780] this time -- a figure -- as 62 showed, the embedding pattern (figure 62 "synchronous patterns") which comprised a synchronous series is repeated twice -- a and To sake, a synchronization can be taken [ ] beforehand -- [ ] -- what kind of -- Timing power only performing processing of every one cycle, even if it starts processing -- someday -- one cycle Since it hits the thing which made the synchronous pattern patrol, a synchronous pattern is detectable.
[0781] At this time, it is a synchronous pattern, for example, shifting one frame at a time in the case of a picture signal etc. Since [ dropping off ] it can process one cycle at a time, the necessity of searching can be efficiently looked for a synchronous pattern.
[0782] a such child -- a figure -- it is shown in 65.
[0783] Digital watermark detecting device all the persons concerned term finishing signal dividing part >
Below, the details of operation of synchronized signal dividing part 740 are explained.
[0784] the amount 3814 of synchronous displacement obtained by synchronous primary detecting element 730 in synchronized signal dividing part 740 -- basis a companion -- the position power with which the synchronization was united is also embedded and it divides one cycle of settled signals 923 at a time.
[0785] Namely, the amount 3814 of synchronous displacement
[0786] [93 number]
<sub>4</sub>taualphatheta
embed by of, and skip settled signal 923 or only the part of T At embeds it -- settled the form which had the embedding pattern and the synchronization when dividing by tracing back signal 923 -- one cycle -- Cut out -- things are made.
[0787] When synchronized partial signal 3817 of the head divided by the detection timing of the synchronous pattern by synchronized signal dividing part 740 is the section for which the synchronous series was embedded in addition, One The case where it is the section where the embedding series of eyes was embedded, and power which may exist About this, it is each division. It is Capacity by trying detection of a synchronous series or an applicable embedding series again to between. Judging to Easy is possible.
[0788] Thus, start the section corresponding to each embedding series as synchronized partial signal 3817, respectively.
[0789] Digital watermark detecting device 1 detection information extraction part >
Below, the example of composition of detection information extraction part 760 is shown.
[0790] a figure -- 66 is an example of composition of the detection information extraction part in the embodiment of The 8 of the present invention.
[0791] Detection information extraction parts 760 are detection series extraction part 761, correlation value calculating part 762, and a maximum judging. It comprises part 763, detection information reconstruction section 764, and detection information connecting part 765, Detection complex butter - A 3815 is inputted, and detection information 3812 is outputted.
[0792] setting to the figure in addition -- a figure -- which composition was indicated that information flows upwards from the bottom, and was noticed about and To in order to make an understanding of correspondence with 60 easy,
[0793] Detection information extraction processing by detection information extraction part 760 is carried out in the following procedures.
[0794] a figure -- 67 is a flow chart of operation of the detection information extraction part in the embodiment of The 8 of the present invention.
[0795] Step 2201 In each part of detection series extraction part 761, correlation value calculating part 762, maximum judgment part 763, and detection information reconstruction section 764, It is a part by the same processing as operation of each part to which detection information extraction part 220 in digital-watermarking detecting device 200 of the embodiment of The 1 corresponds. Part detection information 3615 is acquired.
[0796] However, process each [ which was inputted ] detection complex pattern 3815 of every, respectively, and differ in that detection information reconstruction section 764 outputs a plurality of partial detection information 3615.
[0797] Step 2202 In detection information connecting part 765, a plurality of partial detection information 3615 acquired by detection information reconstruction section 764 is connected, detection information 3812 is constituted, and this is outputted.
[0798] Connection of a plurality of partial detection information 3615 is the embedding information on an electronic watermark embedding device. It has been reverse processing of the split application in dividing part 114. For example, it is the last power of Embed information 911 to embedding information dividing part 114. It is made to be divided into K bits at a time in order, and may be made for To go case to connect partial detection information 3615 in order.
[0799] Example > of composition of others of a Digital watermark detecting device
In the above-mentioned example, it is at synchronized signal dividing part 740, one-cycle [ every ] embedding finishing signal Power which showed the example to divide Electronic Penetration complex arrangement generation part 116 of embedding device 100 -- It is. making it repeat four embedding complex pattern power or more corresponding to a synchronous series -- Fill -- Included -- rare -- ing -- a case -- To -- burying a term every two or more rounds in synchronized signal dividing part 740 even if it divides a signal [ crowded ] -- style trap .
[0800] Finishing of one-cycle [ every ] embedding in embedding finishing signal dividing part 710 in the above-mentioned example Although the example which divides signal 923 was shown, Electronic Penetration Complex arrangement generation of embedding device 100 In part 116, Embedding complex pattern power corresponding to each embedding series The following example When being generated so that it may repeat continuously every two or more times like, embedding finishing Trust after judging the head position of the repetitions in item dividing part 710 -- every [ two or more round term ] even if it divides an embedding finishing signal -- Yo .
[0801] SP, SP, SP, SP, A, A, A, A, - .., A, A, SP, SP, and --
1 1 2 2 k k
(118)
in order to judge the head position of the repetitions -- the amount 3814 of synchronous displacement -- one cycle every section started with one cycle since the starting point was already clear -- a synchronous series -- or -- the -- It is possible to judge easily by trying detection of this embedding series to carry out.
[0802] Feature > of the embodiment of Mr. 8
according to the electronic watermark embedding device and digital-watermarking detecting device of this embodiment -- the time abundant by embedding partial embedding information which is different for every section of a signal by division -- burying -- Included It becomes possible to see and to embed information to a signal.
[0803] Since 1 cycle starting point can be easily matched the synchronization also in digital-watermarking detection, the case of the round robin technique, i.e., a picture signal, of trying the amount of displacement one by one -- said -- Or [ it is not based on the technique of matching and searching for a term signal for every one frame gap but efficient ] One -- matching the synchronization at high speed -- electronic Penetration It becomes possible to detect.
[0804] It may be made to embed as a modification of the electronic watermark embedding device of this embodiment, so that a synchronous series may be superimposed on each embedding series. From a synchronous series to namely, raw The embedding pattern and Addition which are generated from each embedding series in the embedding pattern to accomplish Calculation is carried out and it embeds as follows.
[0805] SP+A, SP+A, SP+A, - .., SP+A, and SP+A --
1 2 3 k 1
Thus, when embedding is performed, it smells at the time of detecting a synchronous series step 2101. It is a synchronous series, without dividing The and embedding finishing signal 923 into the same length as the cycle of a periodic signal. A synchronous series is detectable with sufficient accuracy using the embedding finishing signal of sufficient quantity for detection. this , -- profit -- , -- embedding based on the amount of Was done synchronous displacement, and dividing a settled signal -- each embedding series -- Inspection It is the same as that of what was already described that it can come out and carry out.
[0806] [Embodiment of The 9]
Quadrature conversion field embedding>
Below, the electronic watermark embedding device in the embodiment of The 9 is explained. [0807] Another composition of an electronic watermark embedding device [ in / in this embodiment / the embodiment of The 1 ] An example is shown.
[0808] <Electronic watermark embedding device>
An electronic watermark embedding device and an electron The example of composition of a watermark detecting device is shown. [ in / in Drawing 68 / the embodiment of The 9 of the present invention ]
[0809] Electronic watermark embedding device 800 in this embodiment is complex pattern generation part 810.
Embedding pattern superposed part 820, signal conversion section 830 before embedding, embedding finishing signal contrary It comprises conversion part 840 and storage part 850 of The 1, and they are embedding information 911 and a signal before embedding.
912 is inputted and embedding finishing signal 923 is outputted.
[0810] Below, explain operation of electronic watermark embedding device 800.
[0811] the procedure of the following [ processing / by electronic watermark embedding device 800 / digital-watermarking embedding ] -- real -- It is given.
[0812] a figure -- 69 is a flow chart of operation of the electronic Penetration embedding device in the embodiment of The 9 of the present invention.
[0813] Step 2301 It is Based to embedding information 911 inputted in complex pattern generation part 810! /and Embed complex pattern 4021 are generated and it stores in storage part 850 of The 1.
[0814] Processing of complex pattern generation part 810 is the same as that of complex pattern generation part 110 in electronic watermark embedding device 100 of the embodiment of The 1.
[0815] Input the signal of the length of predetermined section T from signal 912 before step 2302 embedding.
[0816] Every [ of the section obtained at Step 2302 in signal conversion section 830 before step 2303 embedding ] position (X, X, - - -, X) Frequency decomposition is carried out and there is finishing [ a one-dimensional dispersion Fourier transform is carried out, and ] of conversion.
1 2 N- 1
Signal 4022 before embedding is acquired.
[0817] Mention the details of operation of signal conversion section 830 before embedding below.
[0818] In step 2304 embedding pattern superposed part 820, The embedding complex obtained at Step 2301 to signal 4022 before changed embedding acquired at Step 2303 Pattern 4021 is superimposed and embedding finishing signal 4023 before inverse transform is acquired.
[0819] Mention the details of operation of embedding pattern superposed part 820 below.
[0820] It is one-dimensional dispersion in every position (X, X, ..., X) to embedding finishing signal 4023 before inverse transform acquired at Step 2304 in step 2305 embedding finishing signal inverse transforming part 840.
1 2 N-1
Inverse Fourier transform is carried out and embedding finishing signal 923 is acquired.
[0821] Mention the details of operation of embedding finishing signal inverse transforming part 840 below.
[0822] It is the above-mentioned step 230 until it finishes processing signal 912 before step 2306 embedding altogether.
2*2305 is repeated.
[0823] Digital watermark embedding device Signal conversion section > before embedding
Below, the details of operation of signal conversion section 830 before embedding are explained.
[0824] Signal of Section ding taken out from signal 912 before embedding in signal conversion section 830 before embedding One dimension carries out a discrete Fourier transform, and it carries out frequency decomposition.
[0825] Explain below using a formula concretely.
[0826] Signal 912 before embedding It is referred to as iota (chi, chi, ..., chi, t).
1 2 N-1
[0827] Carry out the one-dimensional dispersion Fourier transform of the I (x, x, ..., x, t) as follows, and it is r? (x, x, ..., x)
1 2 N-1 1 2 N u is obtained.
[0828] [94 number]
(1 1 9)
<img file="WO2007102403A1_D0056.tif" />
However, tau presupposes that it is the predetermined number of specimens decided beforehand.
[0829] 7? (chi, X, ..., X, and u) are outputted as signal 4022 before changed embedding.
1 2 N-1
[0830] <electronic watermark embedding device Embedding pattern superposed part >
Below, the details of operation of embedding pattern superposed part 820 are explained.
[0831] With embedding pattern superposed part 820, It was obtained by signal conversion section 830 before embedding. Signal 4022 before changed embedding of nu dimension, It corresponds to specific frequency. nu -- One-dimensional plane part It was generated in complex pattern generation part 810 to the part. nu -- One-dimensional embedding complex putter It superimposes by adding A 4021, The whole containing the frequency of the superimposed result Signal of nu dimension It outputs as front [ inverse transform ] embedding finishing signal 4023.
[0832] Explain below using a formula concretely. [0833] It is r about signal 4022 before changed embedding acquired by signal conversion section 830 before embedding? embedding complex pattern 4021 obtained in (chi, x, - .., chi, u), and complex pattern generation part 810
2 N- 1
It is embedding finishing signal 4023 before inverse transform which is set to rho (chi, X, - - -, X), and is generated 7?' (chi)
1 2 N- 1 1
X It is referred to as - - -, X, and U.
2 N- 1
[0834] [95 number] (H = W)<sub>0</sub>At the time of of <img file="WO2007102403A1_D0057.tif" />+ alpharho * {chi] (w = Facial - u)<sub>0</sub>Time of of eta{chi, chi, -, chi, and upsilon ("!=w)<sub>0</sub>At the time of of
(1 2 0)
however, * expresses complex conjugate -- the frequency u was beforehand decided to be, and U -- the number of frequency specimens
0
Suppose that it is. That is, they are u=u and u=U here. -- N corresponding to frequency u in u -- 1st order
0 0 0
It will be chosen as a Motohira side. u=u and u=U -- It is u and is a conjugate complex number of P.
0 0
That of give is because the signal acquired as a result of discrete inverse Fourier transform serves as a real value.
[0835] It is an intensity parameter, a is constituted so that it may change according to the characteristic quantity with which the whole signal 912 before embedding or part power is also computed, and style trap and the point also of and The are the same as that of the case of embedding pattern superposed part 140 in the embodiment of The 1.
[0836] The 1 also carries out the point which may be added so that embedding complex pattern 4021 may be repeated, when larger [ the size of signal 912 before embedding embeds, and ] than complex pattern 4021. It is the same as that of the case of embedding pattern superposed part 140 in a form.
[0837] Precede superposition of an embedding complex pattern and they are two or more times about embedding complex pattern 4021, Or even if it makes it expand so that it may be equal to the size of signal 912 before embedding, a good ! point is the same as that of the case of embedding pattern superposed part 140 in the embodiment of The 1.
[0838] u=u actually overlapped in advance of superposition of an embedding complex pattern and u=u
0
-It is N about the portion of changed embedding signal 4022 used as u. -- By one-dimensional discrete Fourier transform
0
superimposing, after changing -- further -- N -- inverse transform is carried out by one-dimensional discrete inverse Fourier transform It may carry out for obtaining.
[0839] N -- when performing a one-dimensional discrete Fourier transform, it unites with the one-dimensional dispersion Fourier transform by signal conversion section 830 before embedding -- even if it processes as one discrete Fourier transform of N dimension It is good. Being the same and N -- one-dimensional discrete inverse Fourier transform -- the below-mentioned embedding finishing signal it unites with one-dimensional dispersion inverse Fourier transform in inverse transforming part 840 -- 1 time of N dimension dispersion reverse Huu It may process as Rie conversion. However, it is a one-dimensional dispersion Fourier transform and N--1 as mentioned above. The discrete Fourier transform of a dimension, one-dimensional dispersion inverse Fourier transform, and N -- One-dimensional discrete reverse Fourier u=u and u=U which are actually overlapped by changing separately -- N of u
0 Only One! and The are N to 0 one-dimensional plane. -- A one-dimensional discrete Fourier transform and discrete inverse Fourier transform Since it will end if it carries out, there is an advantage which can perform processing at high speed.
[0840] Embed finishing signal inverse transforming part >
It embeds at below and explains the details of operation of ending signal inverse transforming part 840.
[0841] In embedding finishing signal inverse transforming part 840, embedding finishing signal 4023 before inverse transform is located (X).
One-dimensional dispersion inverse Fourier transform is carried out to X, - - -, and every X, and embedding finishing signal 923 is acquired.
2 N-1
[0842] Explain below using a formula concretely.
[0843] It is embedding finishing signal 4023 before inverse transform? 7 It is considered as '(chi, chi, - - -, x, u).
1 2 N-1
[0844] r? One-dimensional dispersion inverse Fourier transform of '(x, x, - - -, x, u) is carried out as follows, and it is gamma (chi, X, and - - -).
1 2 N-1 1 2
t) [ X and ] obtain.
N-1
[0845] [96 number]<img file="WO2007102403A1_D0058.tif" />
u=0
Feature > of the embodiment of Mr. 9
According to the electronic watermark embedding device of this embodiment, it is Electronic penetration of the embodiment of The 1. Electronic Power which lends and has the same feature as an embedding device can be embedded.
moreover -- setting in complex pattern generation part 810 by the same method as the embodiment of The 7 -- double -- Generate the complex pattern of a number and set to embedding pattern superposed part 820. changed Embedding N corresponding to [ see and ] a plurality of frequency of last signal 4022 -- a one-dimensional plane portion -- each complex pattern It is the same as that of the electronic watermark embedding device of the embodiment of The 7 by making it add. Electronic Penetration with the feature It can embed.
[0847] [Other embodiments]
Combining with below with each embodiment as an embodiment of others of the present invention The made example of composition is shown.
[0848] <Pre filter use at the time of detection>
in each embodiment of the 1st thru/or Z, and The 8, the sine wave was used as a periodic signal a case -- electronic Power -- embedding -- having -- Listen -- a thing -- eye N dimension -- it becomes a single frequency of a direction (in for example, the case of a picture signal the direction of time). They are the foundations also when other periodic signals are used. Frequency is the most important. Finishing of embedding in advance of detection by a digital-watermarking detecting device It is Electronic penetration with high precision by performing filter processing which emphasizes the frequency of relevance to a signal. It is ! even if it can be made to perform detection of Power.
[0849] Use digital filters, such as a FIR filter and an IIR filter, as an example of a filter, and it is Special. The zone passage type filter which emphasizes the frequency band of a law may be constituted. It is considered that the signal value which exceeds or is less than the clipping filter which holds down the signal value which exceeds or is less than a predetermined threshold to an applicable threshold, and a predetermined threshold is 0. A nonlinear filter called epsilon filter etc. It is using, Remove efficiently the noise ingredient for digital watermarking, such as an original picture ingredient. Filter processing which leaves a power S digital-watermarking ingredient may be performed.
[0850] Use a plurality of frequency bands in the embodiment of The 7 of the present invention using a plurality of periodic signals, and it is electronic Penetration. Although embedded, Each periodic signal is received. Filter which has the characteristic united with each periodic signal in advance of processing of a time recovery Even if it uses, respectively and is made to perform filter processing, it is Yo .
[0851] especially -- the present invention in order to perform embedding of electronic Power using the single frequency of the direction of nu dimension eye, there is no straight line phase characteristic -- wrong of the phase characteristic even if it uses Amber filter -- detection performance influence -- Therefore, it is few like an IIR filter instead of wrong ! of the phase characteristic! It is possible to become possible to use the possible filter of high-speed processing with a sharp frequency characteristic with /and the number of taualpharho, and to perform highly precise digital-watermarking detection processing at high speed.
[0852] Processing > to a Embed finishing signal
in each embodiment of the present invention -- for example, a figure -- 10 and a figure -- 52 and a figure -- 68 -- Electronic penetration the embedding finishing signal which carried out and was outputted from the embedding device -- a direct digital-watermarking detecting device -- ON Although it is drawn so that power may be carried out Modification [ an embedding finishing signal / compression, coding, distribution, edit and ] After carrying out When, it cannot be overemphasized that it may be inputted into a digital-watermarking detecting device. It buries and is magnetic media (for example, VTR, DVD) about the signal [ crowded / ]. The account of - And records to the media (film etc.) of a floppy disk (registered trademark), CD, H DD, etc. and others, Transmit through a network, and it Is sufficient, and reproduces using an optical device (for example, it projects on a screen as a movie). They are a video camera and Mobile phone about what was expressed as the display of CRT, a liquid crystal, plasma, etc., etc. It is a style trap even if it re-takes a photograph using the photographing device of the camera of the talk, the camera using a film, etc. It is and things cannot be overemphasized.
[0853] Time abnormal-conditions processing >
In each embodiment of the present invention, it is called for convenience a "time modulation part" and a "time demodulation section". Must which is that in which it not necessarily modulates the direction of a time-axis in an actual signal although it is N of the origin which drops off in short -- It is even if it is the abnormal conditions of the different direction of a dimension if it is a dimension which intersects perpendicularly with one dimension. It does not matter.
[0854] For example, it is electronic Penetration to the image signal with which two-dimensional signal power also becomes. When embedding, It defines as the transverse direction of a picture. It is N about one-dimensional complex arrangement. -- It is a style as a one-dimensional embedding complex pattern. It accomplishes and modulates this in a lengthwise direction. It is a style trap even if it obtains a two-dimensional embedding pattern. It is and is .. It cannot be overemphasized that length and width may be replaced.
[0855] For example, the two-dimensional direction of space (X, Y) and the direction of time It totals by one dimension. When digital watermarking is embedded at a three-dimensional picture signal, It defines as the transverse direction and the direction of time of a picture. A two-dimensional complex It is N about arrangement. -- Even if it constitutes as a one-dimensional embedding complex pattern, this is modulated in a lengthwise direction and it obtains the embedding pattern of 3 dimension, it is style trap . even if it replaces length and width! / -- things It is needless to say.
[0856] in each embodiment of the present invention, the signal before embedding which is an input signal receives the input signal of the power M dimension (>N) which explained to the example the case where it was a signal of N dimension -- N dimension even if it constitutes so that embedding may be repeated -- .
[0857] For example, the two-dimensional direction of space (X, Y) and the direction of time It totals by one dimension. As opposed to the input of a three-dimensional picture signal, Every frame picture of an image It is regarded as a two-dimensional signal and is a transverse direction as mentioned above. One-dimensional complex arrangement was modulated in the lengthwise direction. A two-dimensional embedding pattern is constituted and embedded. It carries out and digital watermarking buries by carrying this out repeatedly to all the frames. It may be crowded. facing detection of electronic Power -- it is processing for every frame -- Go -- even if it processes to and the signal which carried out and superimposed each frame picture -- style trap .
[0858] Use > of Error correction code etc.
In each embodiment of the present invention, it is processing of the embedding information on an embedding series generation part. Even if it precedes, it embeds using an error correcting code and it carries out numerals Y of the information, they are Well, on the contrary Detection information. Even if it decodes an error correcting code in advance of the output of news, it is Yo .
[0859] ToN -- One-dimensional rectangular conversion >
At the embodiment of The 3 of the present invention, it is N. -- One-dimensional inverse Fourier transform part 113, N -- One-dimensional Off N in 1 Rie conversion part 225 -- As the example of the rectangular conversion to a one-dimensional complex pattern Although explained using the discrete Fourier transform, Complex number power other than a discrete Fourier transform Complex number Even if it uses the rectangular converting method which changes, it is style trap .
[0860] Coefficient e which occurred by synchronous displacement of the direction of eye N dimension in order to be, able to process correctly synchronous displacement of the direction of eye N dimension in addition in the field by which rectangular conversion was carried out<sup>jA 9</sup>If it is the conversion Save(ed) Since Well rectangular cross conversion is linear transformation, when this condition is rectangular conversion, it is already Satisfied. It is carried out.
[0861] Linear transformation > of One dimension
As the example of one-dimensional conversion [ in / at the embodiment of The 2 / time modulation part 130b ] Although explained using the discrete Fourier transform, complex number power other than a discrete Fourier transform Complex number it is the linear transformation method of changing and has based on the periodic function which fulfills the following conditions -- as long as it is the linear transformation method that inverse transform exists, what kind of thing may be used.
[0862] It is one-dimensional dispersion Furi as an example of one-dimensional conversion [ in / similarly / at the embodiment of The 9 of the present invention / signal conversion section 830 before embedding, and embedding finishing signal inverse transforming part 840 ]. Although explained using A conversion and one-dimensional dispersion inverse Fourier transform, Except a discrete Fourier transform Complex number power Circumference by which it is the linear transformation method of changing a complex number, and fulfills the following conditions If it is the linear transformation method that have based on a term function and inverse transform exists, anything is. It does not matter. [0863] Conditions:
1) The result with which it integrated by periodic is set to 0.
[0864] It is a peak with two sharp autocorrelation functions Do not have !
[0865] It is as having already given details of the conditions of these as an example of a periodic signal.
[0866] For example, you may be the following linear transformation.
[0867] [97 number]
VectorieC" --;i^Vectore<sup>n</sup>
Linear transformation is considered and the conversion procession showing conversion is set to A. : C is a set of the whole complex number. It expresses.
[0868] [98 number]
(122) It is here,
(1 23) (1 24)
= (yiy<sub>2</sub>--y<sub>n</sub>f
<img file="WO2007102403A1_D0059.tif" />
It carries out.
[0869] At this time f Let (t) be a periodic function of cycle n which fulfills the above-mentioned conditions,
[0870] [99 number]
(126)<img file="WO2007102403A1_D0060.tif" />It is even if it is the linear transformation denoted by becoming conversion procession A. j is an imaginary unit here.
[0871] f as [ showed / in figure 4 A*C / (t) ] (figure 4A) -- they may be the (a) sine wave (figure 4B), the (b) triangular wave (Drawing 4C), and (c) rectangle wave.
[0872] Use > as a Synchronization signal
It is business in the embedding pattern itself which expresses embedding information with the embodiment of The 5 of the present invention. Although how to be and to detect the amount of synchronous displacement of the direction of time was shown, it is the amount detection of synchronous displacement of the present invention. They are arbitrary electronic Penetration about a method. It may use in combination with a method. Namely, arbitrary electrons Penetration It combines with the electronic Penetration embedding by a method, Electronic watermark embedding method of the present invention It is electronic Penetration so that it may use and a matching synchronization signal for exclusive use may be embedded. An embedding device is constituted. It carries out, The amount detecting method of synchronous displacement of the present invention is used, and the matching synchronization signal power is also synchronous Fit. After performing To, it embeds using arbitrary electronic watermark detecting methods, information is detected, and it comes out. A digital-watermarking detecting device may be constituted so that power may be carried out.
[0873] Even if it makes the cycle of the periodic signal for embedding a synchronous series become an integral multiple of the cycle of the periodic signal for embedding an embedding series, it is Yo .
[0874] <Others>
The composition shown by each embodiment of the present invention may be combined suitably, and may be used.
[0875] An above-mentioned electronic watermark embedding device and a digital-watermarking detecting device of each embodiment Operation of each component is built as a program, and it installs in a computer and performs. It is possible to carry out or to make it circulate via a network.
[0876] They are a Node disk and flexible disk'CD about the built program. -- It stores in portable storages, such as ROM, and installs in a computer, or distributes. It is possible.
[0877] According to [ as explained above ] one embodiment of the present invention N (N is an integer greater than or equal to 2) Embed to an input signal with the above dimension, and perceive in man's consciousness by making information into digital watermarking. !/[ have ] to carry out, it is an electronic watermark embedding device embedded like -- the above-mentioned embedding information -- basis a companion -- embedding series creating means which generates an embedding series and it stores in the memory measure of The 1, It is based on the above-mentioned embedding series of the memory measure of above-mentioned The 1, and is N. -- Arrangement which generates a one-dimensional pattern Sequence creating means, The above-mentioned N -- A periodic signal is modulated according to the value on a one-dimensional pattern. Modulation means which generates the embedding pattern of a Ri N dimension and it stores in the memory measure of The 2, front -- it being stored in the memory measure of account The 2, and acquiring the embedding pattern of a The N dimension -- this Fill -- it has an embedding pattern superposing means which superimposes a lump pattern on the above-mentioned input signal The electronic watermark embedding device characterized by things is provided.
[0878] According to this electronic watermark embedding device, it is N. -- It is modulating a one-dimensional pattern in the direction of eye N dimension, and performing embedding, N -- It is N dimension space about the embedding information on a one-dimensional pattern. With the spread redundancy, sufficient tolerance also, for example to change of high compression, re-photography, etc. There and quality degradation can be suppressed and the long information on information length can be embedded as digital watermarking.
[0879] N -- not being concerned with the amount of synchronous displacement in the direction of eye N dimension in one-dimensional space using the embedding series by which spectral diffusion was carried out -- the needlessness of matching synchronization -- or easy -- and -- high-speed -- said -- Possible electronic Power united a term can be embedded.
[0880] The above-mentioned modulation means is the above-mentioned N. -- which may be constituted so that the phase of the direction of eye N dimension may change with positions on a one-dimensional pattern, respectively and the embedding pattern of N dimension may be generated. According to this composition, it is embedding electronic Power using the phase of a periodic signal, N -- It is while being able to modulate the pattern of 1 dimension easily and at high speed in the direction of eye N dimension, Cycle It is not concerned with the amount of synchronous displacement in the direction of eye N dimension using phase displacement of a signal, but is synchronous Fit. Easy and it embeds [ the needlessness of To, or ] possible electronic Power of matching synchronization at high speed. It can do. moreover -- become less than the minimum picture signal quantization value, and electronic Power buries in fact a lump should do -- being able to prevent and a frame occurring and making it into the transmission way of digital watermarking While being able to use the picture signal of The effectively It is Aim about a frame with large amplitude of electronic Power. If Modify is carried out, the tolerance over Attack can be increased when.
[0881] N of an embedding pattern -- The phase is spread on one-dimensional space, the size of the noise ingredient which embeds in as a result of correlation computation, originates in a last signal, and appears -- more -- small -- a fence -- a result -- more reliable electronic Power -- becoming embedding and detectable -- Follow The embedding of electronic Power with less quality degradation and detection are possible with reliability comparable as Come. It becomes.
[0882] It is embedding digital watermarking using the sum of two periodic signals which intersect perpendicularly with the same fundamental frequency, N -- They are abnormal conditions easily and at high speed to the direction of eye N dimension about a one-dimensional pattern. While it is possible, using phase displacement of a periodic signal -- it is a synchronization in the direction of eye N dimension -- strange about -- not being concerned with quantity -- the needlessness of matching synchronization -- or matching synchronization is possible easily and at high speed Digital watermarking can be embedded.
[0883] It has the characteristic which does not have peaks with a sharp autocorrelation function, such as a rectangle wave or a triangular wave, as a periodic signal, using a calculable periodic function easily as compared with a sine wave -- Scarce and environment of and a calculation resource -- ! and The -- embedding processing of digital watermarking -- more -- high-speed -- To realize comes out -- last .
[0884] It is using linear transformation, such as a discrete Fourier transform, for abnormal conditions, For example, high-speed Off 1 Rie conversion etc. are used, and it is N. -- Strange [ in the direction of N dimension / easily and at high speed ] in a one-dimensional pattern While a tone is possible, using linear transformation coefficients, such as a discrete Fourier transform coefficient, -- Nth order not being concerned with the synchronous gap by the direction of a former eye -- the needlessness of matching synchronization -- or easy -- and -- high-speed -- Possible electronic Power of matching synchronization can be embedded.
[0885] It is the !, The, and above-mentioned N to the above-mentioned electronic watermark embedding device. -- A one-dimensional pattern is a complex number pattern, and the above-mentioned arrangement creating means are a real part and a part in a part of above-mentioned embedding series. It is the above-mentioned N so that it may become an imaginary part. -- Even if it generates a one-dimensional pattern, it is Yo .
[0886] While performing embedding using the real part of a complex number, and an imaginary part according to this composition, Putter N in which the restrictions about the symmetry of A drop off -- Embedding is performed using the one-dimensional whole space. It can do, the Le diffuse series length outside Space can be lengthened -- the reliability of detection -- high Tooth and the former It comes out to carry out electronic With power of the longer embedding information length, and to embed it with the reliability of the same grade. coming -- reliability comparable as the former, and information length -- electronic Penetration with less quality degradation It embeds. Things are made.
[0887] It is the !, The, and above-mentioned N to the above-mentioned electronic watermark embedding device. -- A one-dimensional pattern is complex number No turn, and it is the above-mentioned modulation means, The above-mentioned N -- The angle of deviation of the complex number on a one-dimensional pattern The above-mentioned periodic signal is modulated so that it may become a phase of an abnormal-conditions signal and an absolute value may serve as a size of an abnormal-conditions signal. It is A little as it carries out.
[0888] According to this composition, it is not concerned with the amount of synchronous displacement in the direction of eye N dimension, but is the needlessness, or easy and possible electronic Penetration of matching synchronization of matching synchronization at high speed. It can embed. To. Since the angle of deviation of a complex number serves as a phase of the abnormal-conditions signal of the direction of the axis of eye N dimension, Minimum Off into which it becomes less than a picture signal quantization value, and the embedding of electronic Power is not made in fact It can prevent Rehm's occurring and effective in the picture signal as a transmission way of electronic Power While being able to use, When it changes targeting a frame with large amplitude of electronic Power, it is Attack when. The tolerance over Shock can be increased.
[0889] In the above-mentioned electronic watermark embedding device, it is the above-mentioned embedding series creating means, The generated above-mentioned embedding series is divided, a plurality of embedding series are generated, and it is an account of above-mentioned The 1. It stores in Means and is the above-mentioned arrangement creating means, the above stored in the memory measure of above-mentioned The 1 -- double -- N respectively corresponding for every embedding series of a number -- generating a one-dimensional pattern -- the above-mentioned modulation means Is The above-mentioned N -- The embedding pattern of N dimension respectively corresponding for every one-dimensional pattern is generated, and it stores in the memory measure of and above-mentioned The 2, and is the above-mentioned embedding pattern superposing means, account of above-mentioned The 2 Yo superimposed on the above-mentioned input signal after adding all the above-mentioned embedding patterns of Means It may constitute to obtain.
[0890] According to this composition, it is embedding information using a plurality of periodic signals, Electronic With power of the embedding information that information length is longer can be carried out, and it can be embedded, and reliability of a detection result can be clarified more correctly, and spectral diffusion series length can be lengthened and high of reliability and electronic Penetration embedding can be performed more.
[0891] moreover -- according to one embodiment of the present invention -- N -- it embeds to an input signal with the above (N is integer greater than or equal to 2) dimension, electronic With power of the information is carried out, and it is not detected by man's consciousness -- as -- Electronic Penetration to embed It is an embedding device. It embeds based on the above-mentioned embedding information. Embedding series creating means which generates a series and it stores in the 1st memory measure, Account of above-mentioned The 1 It is based on the above-mentioned embedding series stored in Means, and is N. -- Arrangement creating means which generates a one-dimensional pattern and it stores in the 2nd memory measure, Rectangular conversion of the above-mentioned input signal is carried out, and there is finishing of conversion. The above-mentioned N stored in the conversion method which acquires a signal, and the memory measure of above-mentioned The 2 -- They are a part of N of the above-mentioned changed signal about One-dimensional butter - A. -- It superimposes on a primary plane, the signal before inverse transform is acquired -- burying carrying out rectangular inverse transform of the above-mentioned signal before inverse transform to a lump pattern superposing means -- embedding finishing signal The electronic watermark embedding device having an inverse transform means to acquire is provided. To. [0892] According to this electronic watermark embedding device, it is N. -- It is the direction of eye N dimension about a one-dimensional pattern. It is embedding as a signal, superimposing on a last signal, and performing embedding, N -- One-dimensional putter By the redundancy which spread the embedding information on A to N dimension space, they are high compression, re-photography, etc., for example. It is Electronic penetration about the information which there is sufficient tolerance also to change and suppresses quality degradation that information length is long. Force can be carried out and it can embed. There are no restrictions about the symmetry of a pattern and it is N--1. Embedding can be performed using the whole space of a dimension, longer with the reliability with the reliability of detection comparable as high Tooth and the former which can lengthen spectral diffusion series length -- burying -- Included it sees, and electronic With power of the information length can be carried out, and it can be embedded -- reliability comparable as the former, and information length Electronic Power with less quality degradation can be embedded. N -- One-dimensional empty The embedding series by which spectral diffusion was carried out in between is used, and it is the amount of synchronous displacement in the direction of eye N dimension. It is not concerned but they are the needlessness of matching synchronization, or easy and the possible electron of matching synchronization at high speed. Penetration It can embed. moreover -- becoming less than the minimum picture signal quantization value -- actually -- To -- it can prevent the frame which embedding of electronic Power is not made occurring -- electron While being able to use the picture signal as a transmission way of Power effectively the amplitude of digital watermarking -- a large -- Like -- if it changes targeting a frame, the tolerance over Attack can be increased when.
[0893] They are !, The, and the above-mentioned embedding series creating means to the above-mentioned electronic watermark embedding device, A plurality of embedding series are generated, and it stores in the memory measure of above-mentioned The 1, and is the above-mentioned arrangement creating means, Before N one-dimensional pattern respectively corresponding for every above-mentioned embedding series of a plurality of stored in the memory measure of account The 1 is generated, storing in the memory measure of above-mentioned The 2 -- the above-mentioned embedding pattern The above-mentioned N by which the superposing means is stored in the memory measure of above-mentioned The 2 -- it is the above about a one-dimensional pattern. A plurality of N of a changed signal -- it superimposes on a one-dimensional plane, respectively -- even if it makes it like -- Yo .
[0894] A plurality of N of a changed signal [ according to this composition ] -- It is N to a one-dimensional plane. -- It is superimposing a one-dimensional pattern, This which embeds merit ! and the embedding information on information length as digital watermarking more It can do, reliability of a detection result can be clarified more correctly, and it is A. Vector diffuse series length can be lengthened and they are high of reliability, and digital-watermarking embedding more. It can carry out.
[0895] According to one embodiment of the present invention, it is N. Electronic Penetration embedded so that it might not be beforehand detected by man's consciousness to an input signal with the above (N is integer greater than or equal to 2) dimension Inspection It is a digital-watermarking detecting device come out of and carried out, Place in the one direction of a dimension of the above-mentioned input signal The ingredient of the periodic signal of a law is measured and it is N. -- Demodulation means which asks for a one-dimensional pattern, The above-mentioned N -- One A detection series extraction means to calculate a detection series from the value of a dimension pattern, and to store in a memory measure, It embeds with the above-mentioned detection series stored in the above-mentioned memory measure, and is Based to the size of the correlation value of a series. It is Special about having a correlation value calculating means which is and detects electronic Power currently embedded. The digital-watermarking detecting device considered as the mark is provided.
[0896] According to this digital-watermarking detecting device, it is N. -- Enough [ as opposed to / for example, / change of high compression, re-photography, etc. ] by the redundancy which spread the embedding information on a one-dimensional pattern to N dimension space are . It is tolerant, quality degradation is suppressed and detection of electronic Power can do the long information on information length. Also and N -- using the embedding series by which spectral diffusion was carried out in one-dimensional space -- eye N dimension not being concerned with the amount of synchronous displacement in a direction -- the needlessness of matching synchronization -- or easy -- and -- high-speed -- synchronization Detection of possible electronic Power to unite can be performed.
[0897] It is using linear transformation, such as a discrete Fourier transform, for a recovery, using for example, high-speed Off 1 Rie conversion etc. -- Signal power N of N dimension -- easy in a one-dimensional pattern -- and -- high-speed -- While being able to get over using linear transformation coefficients, such as a discrete Fourier transform coefficient, -- Nth order not being concerned with the synchronous gap by the direction of a former eye -- the needlessness of matching synchronization -- or easy -- and -- high-speed -- Detection of possible electronic Power of matching synchronization can be performed.
[0898] The above-mentioned demodulation means generates two periodic signals with the same frequency which intersect perpendicularly, is based on correlation with the above-mentioned input signal and the above-mentioned periodic signal, and is N. -- It is composition so that it may ask for a one-dimensional pattern. It carries out and Chi is good.
[0899] According to this composition, it is N. -- It is while being able to restore to a one-dimensional pattern easily and at high speed in the direction of eye N dimension, using phase displacement of a periodic signal -- it is a synchronization in the direction of eye N dimension not being concerned with the amount of displacement -- the needlessness of matching synchronization -- or matching synchronization is possible easily and at high speed are . Digital-watermarking detection is possible.
[0900] It has the characteristic which does not have peaks with a sharp autocorrelation function, such as a rectangle wave or a triangular wave, as a periodic signal, This in which and a calculation resource Scarce by using a calculable periodic function easily as compared with a sine wave, and , environmental Well , and The also realize detection processing of digital watermarking at high speed coming out -- last . [0901] The above-mentioned demodulation means is the difference or differential value of a direction of eye N dimension about the above-mentioned input signal. It can constitute so that machine One and Demodulation may be performed.
[0902] According to this composition, the digital-watermarking detection with high detecting accuracy is attained by performing the recovery by a periodic signal using the difference or differentiation of a signal, and they are V, fewer than that of Signal degradation, and electronic Penetration to comparable detection performance. A method is made possible.
[0903] In the above-mentioned digital-watermarking detecting device, it is the above-mentioned N. -- A one-dimensional pattern is a complex number pattern, and the above-mentioned detection series extraction means is the above-mentioned N. -- To the value of the real part of a one-dimensional pattern, and an imaginary part It is good also as being based, searching for the above-mentioned detection series, and storing in the above-mentioned memory measure.
[0904] There are no restrictions about the symmetry of a pattern by detection of digital watermarking which was embedded using the real part of a complex number, and the imaginary part according to this composition, and it is N. -- It buries using the one-dimensional whole space. Since it is crowded, This from which spectral diffusion series length can be lengthened, and the reliability of detection carries out electronic With power of the longer embedding information, and detects it with high reliability comparable as and the former It can do and is Inspection of electronic Power with less quality degradation at reliability comparable as the former, and information length. Appearance becomes possible.
[0905] In the above-mentioned digital-watermarking detecting device, it is the above-mentioned N. -- A one-dimensional pattern is complex number Bataan, The above-mentioned correlation value calculating means calculates the complex correlation value for every bit, and is every bit. After arranging direction of a complex correlation value, those total is taken, and they are Based and Embed to this total. It is good also as detecting rare ing electronic Power. According to this composition, it is for every bit. A bit judging error decreases and it can realize higher tolerance rather than detecting.
[0906] In the above-mentioned digital-watermarking detecting device, it is the above-mentioned N. -- A one-dimensional pattern is complex number Bataan, the above-mentioned correlation value calculating means is buried based on the absolute value of a complex correlation value, and it is Included. It is ! also as detecting Have been and Electronic transmission.
[0907] according to this composition, by detecting using the absolute value of a complex correlation value, it can embed also to the input signal from which the synchronization shifted in the direction of eye N dimension, and can acquire correlation with a series -- said -- Unnecessary digital-watermarking detection united a term can be performed.
[0908] It is the !, The, and above-mentioned N to the above-mentioned electronic watermark embedding device. -- A one-dimensional pattern is a complex number pattern, The above-mentioned correlation value calculating means calculates the complex correlation value for every bit, and is each Bit. After arranging direction of the complex correlation value for every A, those total is taken, and it may be made to have a synchronous means which calculates the amount of synchronous displacement of and the above-mentioned input signal based on the angle of deviation of this total. To this Digital watermarking can measure Ri and the amount of synchronous displacement and according to easy and high-speed matching synchronization It becomes detectable. The amount of synchronous displacement is calculated based on the complex correlation value for every bit. The measurement of the more certain and high-precision amount of synchronous displacement also of Ri is attained.
[0909] It is the !, The, and above-mentioned N to the above-mentioned electronic watermark embedding device. -- A one-dimensional pattern is complex number No turn, and it is Based to the angle of deviation of the complex correlation value of the above-mentioned detection series and the above-mentioned embedding series. It is good also as having a synchronous means which comes and calculates the amount of synchronous displacement of the above-mentioned input signal.
[0910] it comes out to embed by detecting using the angle of deviation of a complex correlation value also to the input signal from which the synchronization shifted in the direction of eye N dimension, to be able to acquire correlation with a series, and to measure the amount of synchronous displacement It comes and the digital-watermarking detection by easy and high-speed matching synchronization is attained.
[0911] In the above-mentioned electronic watermark embedding device, it is the above-mentioned demodulation means, About that of a plurality of periodic signals A phase is measured and they are a plurality of N. -- It asks for a one-dimensional pattern and is the above-mentioned synchronous means, A plurality of N -- One dimension The amount of synchronous displacement is respectively calculated for every pattern, and it is the above-mentioned detection series extraction means, A plurality of above-mentioned N -- Based on the respectively corresponding above-mentioned amount of synchronous displacement, a synchronization is amended and detected from a one-dimensional pattern. A series is searched for, and it stores in the above-mentioned memory measure, and is the above-mentioned correlation value calculating means, A plurality of above-mentioned N -- The above-mentioned detection series stored in the above-mentioned memory measure acquired for every one-dimensional pattern was combined. It is good also as calculating the correlation value of a series and the above-mentioned embedding series.
[0912] According to this composition, it is detecting digital watermarking embedded using a plurality of periodic signals, Electronic With power of the embedding information that information length is longer can be carried out, it can be detected, and it is Up. Reliability of It was and a detection result can be clarified more correctly, and it is spectral diffusion. Series length can be lengthened and it becomes detectable [ more reliable electronic Power ].
[0913] A plurality of [ synchronous means / above-mentioned ] above-mentioned N -- Based on the amount of synchronous displacement obtained for every one-dimensional pattern, it may constitute so that the amount of displacement of the whole direction of an axis of eye N dimension may be calculated. This composition If it depends, the amount of synchronous displacement of each detection complex pattern can be calculated with sufficient accuracy, and it is detecting accuracy. In high and comparable detecting accuracy, little digital-watermarking detection of quality degradation is possible. It becomes ability.
[0914] The above-mentioned synchronous means detects the synchronous series embedded beforehand, it matches the synchronization, re-divides an input signal according to the above-mentioned amount of synchronous displacement, and detects a plurality of embedding information which remains. It is good also as having a detection means to carry out. According to this composition, it was divided in the direction of time. By detecting embedding information, longer embedding information can be embedded to a signal and the application application of and electronic Power can be expanded.
[0915] moreover -- according to one embodiment of the present invention -- N -- it embeds to an input signal with the above (N is integer greater than or equal to 2) dimension, electronic With power of the information is carried out, and it is not detected by man's consciousness -- as -- it is a digital-watermarking embedding program for embedding -- computer, being the above-mentioned -- burying Embedding system which generates an embedding series based on lump information, and it stores in the memory measure of The 1 Sequence creating means, It is based on the above-mentioned embedding series of the 1st above-mentioned memory measure, and is N. -- One-dimensional pattern Arrangement creating means to generate, The above-mentioned N -- A periodic signal is modulated according to the value on a one-dimensional pattern. The embedding pattern of N dimension is generated by carrying out, Abnormal conditions stored in the memory measure of The 2 It is stored in a means and the memory measure of above-mentioned The 2, and the embedding pattern of a The N dimension is acquired. Embedding pattern superposing means which carries out and superimposes the embedding pattern on the above-mentioned input signal Electronic Penetration [ making it function by carrying out ] An embedding program is provided.
[0916] moreover -- according to one embodiment of the present invention -- N -- it embeds to an input signal with the above (N is integer greater than or equal to 2) dimension, electronic With power of the information is carried out, and it is not detected by man's consciousness -- as -- it is a digital-watermarking embedding program for embedding -- computer, being the above-mentioned -- burying Embedding system which generates an embedding series based on lump information, and it stores in the memory measure of The 1 Sequence creating means, N one-dimensional pattern is generated based on the above-mentioned embedding series stored in the 1st above-mentioned memory measure, The arrangement creating means, the above-mentioned Input signal which are stored in the 2nd memory measure Rectangular conversion of the item is carried out, It is stored in the conversion method which acquires a changed signal, and the memory measure of above-mentioned The 2. TheN -- They are a part of N of the above-mentioned changed signal about a one-dimensional pattern. -- It superimposes on a primary plane, reverse -- strange rectangular inverse transform of the embedding pattern superposing means which acquires the signal before Conversion, and the above-mentioned signal before inverse transform is carried out -- digital-watermarking embedding plog which considers it as an inverse transform means to acquire an embedding finishing signal, and it operates Rum is provided.
[0917] According to one embodiment of the present invention furthermore, it is N. Electronic Penetration embedded so that it might not be beforehand detected by man's consciousness to an input signal with the above (N is integer greater than or equal to 2) dimension Inspection It is a digital-watermarking detection program come out of and carried out, and is a computer, One of the above-mentioned input signals The ingredient of the predetermined periodic signal in the direction of a dimension is measured, N -- Recovery which asks for a one-dimensional pattern A tone means, the above-mentioned N -- A detection series is calculated from the value of a one-dimensional pattern, Inspection stored in a memory measure It embeds with the above-mentioned detection series stored in the appearance series extraction means and the above-mentioned memory measure, and is correlation of a series. The digital-watermarking detection program which considers it as the correlation value calculating means which detects electronic Power currently embedded based on the size of a value, and it operates is provided.
Industrial applicability
[0918] The present invention is applicable to the art which embeds digital watermarking at still picture and video, and the art of detecting digital watermarking.
[0919] limiting [ the present invention / to the above-mentioned embodiment ]-in addition Without -- a claim -- inside
V, Various change , and application are possible.
[0920] Japan Patent application 2006 which applied for application on March 7, 2006 at the its native country time -- 061745 The right of priority based on an item is claimed, and the contents of all the are used for application in the their native country case.
Contents54
130 sheets
Sheet 1 Sheet 2 Sheet 3 Sheet 4 Sheet 5 Sheet 6 Sheet 7 Sheet 8 Sheet 9 Sheet 10 Sheet 11 Sheet 12 Sheet 13 Sheet 14 Sheet 15 Sheet 16 Sheet 17 Sheet 18 Sheet 19 Sheet 20 Sheet 21 Sheet 22 Sheet 23 Sheet 24 Sheet 25 Sheet 26 Sheet 27 Sheet 28 Sheet 29 Sheet 30 Sheet 31 Sheet 32 Sheet 33 Sheet 34 Sheet 35 Sheet 36 Sheet 37 Sheet 38 Sheet 39 Sheet 40 Sheet 41 Sheet 42 Sheet 43 Sheet 44 Sheet 45 Sheet 46 Sheet 47 Sheet 48 Sheet 49 Sheet 50 Sheet 51 Sheet 52 Sheet 53 Sheet 54 Sheet 55 Sheet 56 Sheet 57 Sheet 58 Sheet 59 Sheet 60 Sheet 61 Sheet 62 Sheet 63 Sheet 64 Sheet 65 Sheet 66 Sheet 67 Sheet 68 Sheet 69 Sheet 70 Sheet 71 Sheet 72 Sheet 73 Sheet 74 Sheet 75 Sheet 76 Sheet 77 Sheet 78 Sheet 79 Sheet 80 Sheet 81 Sheet 82 Sheet 83 Sheet 84 Sheet 85 Sheet 86 Sheet 87 Sheet 88 Sheet 89 Sheet 90 Sheet 91 Sheet 92 Sheet 93 Sheet 94 Sheet 95 Sheet 96 Sheet 97 Sheet 98 Sheet 99 Sheet 100 Sheet 101 Sheet 102 Sheet 103 Sheet 104 Sheet 105 Sheet 106 Sheet 107 Sheet 108 Sheet 109 Sheet 110 Sheet 111 Sheet 112 Sheet 113 Sheet 114 Sheet 115 Sheet 116 Sheet 117 Sheet 118 Sheet 119 Sheet 120 Sheet 121 Sheet 122 Sheet 123 Sheet 124 Sheet 125 Sheet 126 Sheet 127 Sheet 128 Sheet 129 Sheet 130
Every citation, both waysCites: the store holds 13 of 14
| Document | Relation | Office | Category | Cited during |
|---|---|---|---|---|
| JP2012503930A | Cited by | Japan | – | Examiner |
| US9582842B2 | Cited by | United States of America | – | Applicant |
| WO2012127611A1 | Cited by | World Intellectual Property Organization (WIPO) | – | International search |
| US9607348B2 | Cited by | United States of America | – | Applicant |
| JPWO2012127611A1 | Cited by | Japan | – | Search report |
| US8942413B2 | Cited by | United States of America | – | Applicant |
| JP5729464B2 | Cited by | Japan | – | Search report |
| US8995706B2 | Cited by | United States of America | – | Applicant |
| US8861781B2 | Cited by | United States of America | – | Applicant |
| JP2013168789A | Cited by | Japan | – | Examiner |
| US8712094B2 | Cited by | United States of America | – | Applicant |
| KR20110047106A | Cited by | Republic of Korea | – | Search report |
| JP2014022972A | Cited by | Japan | – | Examiner |
| JP2010278861A | Cited by | Japan | – | Examiner |
| JP2016048851A | Cited by | Japan | – | Search report |
| JP2010045453A | Cited by | Japan | – | Search report |
| JP2000287073A | Cites | Japan | A | Search report |
| JP2000287073A | Cites | Japan | A | International search |
| JP2002290936A | Cites | Japan | A | Search report |
| JP2002290936A | Cites | Japan | A | International search |
| JP2004172758A | Cites | Japan | A | Search report |
| JP2004172758A | Cites | Japan | A | International search |
| JPH0646392A | Cites | Japan | A | Search report |
| JPH0646392A | Cites | Japan | A | International search |
| JPH09191394A | Cites | Japan | X | Search report |
| JPH09191394A | Cites | Japan | X | International search |
| JPH10285562A | Cites | Japan | A | Search report |
| JPH10285562A | Cites | Japan | A | International search |
| JPH11341452A | Cites | Japan | X | International search |
| YAMAMOTO S. ET AL.: "Frame Chojogata Dogazo Denshi Sukaishi no Kenshutsu Seino Hyoka ni Kansuru Ichikosatsu", FIT2005 DAI 4 KAI FORUM ON INFORMATION TECHNOLOGY, 22 August 2005 (2005-08-22), pages 243 - 244 + ABSTR. NO. J-029, XP003017677 | Non-patent | – | – | International search |
| See also references of EP 1993290A4 | Non-patent | – | – | International search |
19 members in 7 offices
Priority claims4
| Document | Office | Kind | Date |
|---|---|---|---|
| 2006061745 | Japan | A | |
| 2006061745 | Japan | A | |
| 2006061745 | – | – | – |
| JP20060061745 | – | – | – |
Members19
| Document | Office | Kind | |
|---|---|---|---|
| WO2007102403A1This record | World Intellectual Property Organization (WIPO) | A1 | |
| TW200822754A | Taiwan Province of China | A | |
| KR20080100244A | Republic of Korea | A | |
| EP1993290A1 | European Patent Office (EPO) | A1 | |
| US2009074242A1 | United States of America | A1 | |
| CN101401429A | China | A | |
| JPWO2007102403A1 | Japan | A1 | |
| KR100975048B1 | Republic of Korea | B1 | |
| JP2011199892A | Japan | A | |
| JP4879968B2 | Japan | B2 | |
| US2012070032A1 | United States of America | A1 | |
| US2012114168A1 | United States of America | A1 | |
| US8208683B2 | United States of America | B2 | |
| CN101401429B | China | B | |
| TWI370688B | Taiwan Province of China | B | |
| JP5235224B2 | Japan | B2 | |
| US8605939B2 | United States of America | B2 | |
| US8620018B2 | United States of America | B2 | |
| EP1993290A4 | European Patent Office (EPO) | A4 |
7 legal events, as 2 offices reported them to INPADOC
Over the term
Point at a mark for the eventEvents
| Event | Code | Office | |
|---|---|---|---|
| Wipo information: entry into national phaseWWE | WWE | WO | |
| Non-entry into the national phaseNENP | NENP | DE | |
| Wipo information: entry into national phaseWWE | WWE | WO | |
| Wipo information: entry into national phaseWWE | WWE | WO | |
| Wipo information: entry into national phaseWWE | WWE | WO | |
| Wipo information: entry into national phaseWWE | WWE | WO | |
| Ep: the epo has been informed by wipo that ep was designated in this application121 | 121 | WO |
Numbers
- Publication
- 2007/102403
- Publication, DOCDB
- 2007102403
- Publication, EPODOC
- WO2007102403
- Application
- 53953
- Application, DOCDB
- 2007053953
- Application, EPODOC
- WO2007JP53953
Titles2
- English
- ELECTRONIC WATERMARK EMBEDDING METHOD, DEVICE, AND PROGRAM, AND ELECTRONIC WATERMARK DETECTING METHOD, DEVICE, AND PROGRAM
- French
- PROCÉDÉ, DISPOSITIF ET LOGICIEL D'INCRUSTATION DE FILIGRANE ÉLECTRONIQUE ET PROCÉDÉ, DISPOSITIF ET LOGICIEL DE DÉTECTION DE FILIGRANE ÉLECTRONIQUE
Classification
- CPC, 7
- H04N19/467
- H04N1/387
- H04N21/23892
- H04N21/44008
- H04N21/4627
- H04N21/8358
- H04N5/913
- IPC, 8
- H04N19 00
- H04N1 387
- H04N7 08
- H04N7 24
- H04N19 196
- H04N19 467
- H04N19 60
- H04N19 70
Designated states4
- Regional, 4
- Zimbabwe
- Turkmenistan
- Türkiye
- Togo