Image processing apparatus and method

0001The present invention relatres to an image processing method and device.

0002Visual contents, be they still or moving images, are in general creations which benefit from copyright-related exclusivity guarantees. Their reproduction is in general permitted only within a strictly defined framework which allows for remuneration of authors and their beneficiaries.

0003In order to ensure that these legal rules are correctly adhered to, numerous systems have been developed to prevent illegal copies or sufficiently impair their quality as to render them unusable.

0004Within this framework, <patcit id="pcit0001" dnum="EP1237369A"><text>patent application EP 1 237 369</text></patcit> aims to combat the copying of images by picture-taking during their display, for example with a camcorder in a cinema auditorium. With this aim, it is proposed that the intensity of the pixels of a pattern be modulated about the value to be displayed at a high frequency which renders the pattern invisible to the human eye but which generates artefacts on the sequence filmed by the camcorder. This pattern is commonly called watermarking or anti-copy pattern. The shape of the pattern is determined so as to inscribe for example messages of the type "ILLEGAL COPY" which will appear in the images displayed by the camcorder.

0005In order for the pattern to be invisible to the naked eye, the modulation consists in alternating images in which the pattern is bright with images in which it is dark, the mean intensity of the pattern over several images corresponding to that to be displayed in the images in the absence of a pattern. During the display of these images, the eye carries out an integration and in fact perceives the mean intensity. This technique may also be applied to the colour of the images by alternating images in which the pattern is more coloured with images in which it is less so, the mean colour of the pattern over several images corresponding to that to be displaced in the absence of a pattern.

0006In practice, the patterns are contained in a modulation image and the images of the sequence to be protected are modulated with this modulation image.

0007One of the principal difficulties of this type of technique is to determine the locations at which to place the pattern or patterns in the images so as to maximize the hindrance to a person watching an illicit copy generated for example by a camcorder.

0008According to the invention, a solution making it possible to solve this difficulty is proposed. According to the invention, the position and/or the shape of the patterns is defined as a function of the source image to be modulated. More particularly, the modulation image is generated on the basis of the source image that it modulates. Preferably, the modulation image is produced on the basis of the secondary contours of the image to be modulated. The secondary contours of the image to be modulated are employed to form the patterns of the modulation image. The patterns therefore get placed in the low-gradient zones of the source image.

0009The invention relates to a method of image processing intended to temporally modulate a sequence of source images with a modulation image comprising at least one pattern, said sequence of source images being modulated with the modulation image in such a way that said at least one pattern is invisible to the human eye when the images of the sequence are displayed and that artefacts appear when said images are copied by picture-taking during their display, characterized in that the shape and/or the position of said pattern of the modulation image is determined by analysis of the source image that said modulation image modulates.

0010More particularly, the modulation image is generated directly on the basis of the source image that it modulates and the modulation image is an image of the secondary contours of the source image that it modulates.

0011The invention will be better understood on reading the description which follows, given by way of nonlimiting example, and with reference to the appended figures among which : <ul id="ul0001" list-style="dash" compact="compact"><li><figref idref="f0001">Figure 1</figref> is a flowchart summarizing the steps for generating a modulation image according to the invention ; and</li><li><figref idref="f0002">Figure 2</figref> illustrates the application of the various steps of <figref idref="f0001">Figure 1</figref> to a source image.</li></ul>

0012According to the invention, each modulation image is calculated on the basis of the image that it modulates. The geometry of the modulation image therefore repeats that of the image that it modulates. Moreover, the processing applied to the source image to generate the modulation image is defined such that the patterns of the modulation image are located in low-gradient zones of the image to be modulated. For this purpose, the image of the secondary contours of the image to be modulated is chosen for use as modulation image.

0013With reference to <figref idref="f0001">Figure 1</figref>, the modulation image is for example generated by applying the following steps to the image to be modulated : <ul id="ul0002" list-style="dash" compact="compact"><li>detection of the contours 100 in the image to be modulated;</li><li>smoothing 200 of the detected contours ; and</li><li>extraction of the secondary contours 300 from the smoothed image.</li></ul>

0014The image of the secondary contours that are thus obtained will then be used as modulation image.

0015The detection of the contours 100 may be done by parallel application of two filterings on the source image to be modulated : a positive filtering with kernel<maths id="math0001"><math display="inline"><mfenced open="[" close="]"><mtable><mtr><mtd><mo>-</mo><mn>1</mn></mtd><mtd><mo>-</mo><mn>1</mn></mtd><mtd><mo>-</mo><mn>1</mn></mtd></mtr><mtr><mtd><mo>-</mo><mn>1</mn></mtd><mtd><mn>8</mn></mtd><mtd><mo>-</mo><mn>1</mn></mtd></mtr><mtr><mtd><mo>-</mo><mn>1</mn></mtd><mtd><mo>-</mo><mn>1</mn></mtd><mtd><mo>-</mo><mn>1</mn></mtd></mtr></mtable></mfenced></math><img file="EP1592243B1_D0001.tif" /></maths> detecting the contours corresponding to light to dark transitions and a negative filtering with kernel<maths id="math0002"><math display="inline"><mfenced open="[" close="]"><mtable><mtr><mtd><mn>1</mn></mtd><mtd><mn>1</mn></mtd><mtd><mn>1</mn></mtd></mtr><mtr><mtd><mn>1</mn></mtd><mtd><mo>-</mo><mn>8</mn></mtd><mtd><mn>1</mn></mtd></mtr><mtr><mtd><mn>1</mn></mtd><mtd><mn>1</mn></mtd><mtd><mn>1</mn></mtd></mtr></mtable></mfenced></math><img file="EP1592243B1_D0002.tif" /></maths> detecting the contours corresponding to dark to light transitions. The result of these two filters is added up to form a contour image. <figref idref="f0002">Figure 2</figref> shows, for a given source image, the result at the end of each of steps 100, 200 and 300. In the image illustrating the detection of the contours in the source image, the contours detected are represented by the most luminous pixels of the image.

0016Preferably, the contours detected are thereafter smoothed to spatially merge the positive contours and the negative contours and to eliminate noise. Specifically, in the case of a fine light band on a dark background, the above filters detect two very closely spaced contours. This smoothing operation 200 makes it possible to merge them. It is for example carried out by a 5x5 averaging operation. Each pixel of the contour image is thus assigned the mean level of the 25 pixels of the block of pixels encompassing the pixel considered, the pixel considered being the centre of said block.

0017In the image of <figref idref="f0002">Figure 2</figref> where the contours have been smoothed, the secondary contours are represented by pixels having a lesser intensity with respect to the most luminous pixels representing the main contours. The secondary contours may then be extracted from this image using a table of LUT (Look Up Table) type favouring the low levels with respect to the high levels. The very low levels are however removed so as to eliminate the contours due to noise. The LUT table is for example defined as follows: <tables id="tabl0001" num="0001"><table frame="all"><tgroup cols="2"><colspec colnum="1" colname="col1" colwidth="31mm" /><colspec colnum="2" colname="col2" colwidth="23mm" /><thead valign="top"><row><entry><b>Input level</b></entry><entry><b>Output level</b></entry></row></thead><tbody><row><entry>From 0 to 23</entry><entry>0</entry></row><row><entry>From 24 to 31</entry><entry>255</entry></row><row><entry>From 32 to 95 (= x)</entry><entry>255-4.(x-32)</entry></row><row><entry>From 96 to 255</entry><entry>0</entry></row></tbody></tgroup></table></tables>

0018The image of the secondary contours which is thus obtained is used as modulation image in the modulation processes aimed at protecting the visual contents against copying by a camcorder. The last image of <figref idref="f0002">Figure 2</figref> represents the resulting image after modulation of the source image by the image of the secondary contours. The spots appearing on the jacket are the result of the processing according to the invention. These spots are yet more visible in the case of a colour image sequence. They deform progressively as a function of the modifications of the composition of the scene. After visualization by a camcorder, these spots flicker or change colour as a function of the display frequency of the display panel used and of the acquisition frequency of the picture-taking device (camcorder).

0019The method may easily be implemented in a conventional digital processor.

0020In conclusion, the method of the invention has the following advantages : <ul id="ul0003" list-style="dash" compact="compact"><li>the patterns generated are not superimposed on the main contours of the source image ; there is therefore approximately a doubling of the number of contours in the image acquired with a camcorder, half of these contours corresponding to nuisance information ;</li><li>the patterns of the modulation image have the feature of following the motions present in the video sequence, and this has made it possible not to create defects related to the motion of the eye during the visualization of the sequence;</li><li>when a contour is present in two images only, the eye can perceive the appearance or the disappearance of the modulation since the eye is very sensitive to the abrupt transitions in the image sequence and since the patterns are placed in the conventional case independently of the content of the image to be modulated ; in the present case, this appearance or disappearance of the modulation is less perceptible to the eye since, during the display of the image sequence, it concentrates on the main contours so as to analyze the image to the detriment of the intermediate zones which exhibit few contours visible to the naked eye.</li></ul>