Motion picture encoder, and program

The present invention relates to a moving image coding device that performs in-field predictive coding and inter-field predictive coding of a moving image composed of fields.

Since the amount of moving image data is generally large, high-efficiency coding is performed when the moving image data is transmitted from the transmitting device to the receiving device, or when it is stored in the storage device. Here, "high-efficiency coding" is a coding process for converting a certain data string into another data string, and refers to a process for compressing the amount of the data.

As a typical moving image coding method, ISO / IEC MPEG-2 / MPEG-4 (hereinafter referred to as MPEG-2, MPEG-4) can be mentioned. When performing real-time communication using such a moving image coding method, it is necessary to shorten the time from coding to reproduction on the receiving side, and coding on the transmitting side and decoding on the receiving side. It is necessary to realize low delay to reduce the delay in.

The MPEG-2 low delay encoding method based on Patent Document 1 will be described below. MPEG-2 defines three types of pictures, I, P, and B. The I-picture is a coded image of another picture and a picture in which the image can be restored only by the information in the own picture without using prediction (in-frame coding). The P-picture is a picture in which prediction is performed in the forward direction from a past picture and the prediction error is encoded (inter-frame coding). The B picture is a picture in which bidirectional inter-picture prediction is performed from past and future pictures, and the prediction error is encoded (inter-frame coding). Since the B picture uses the future picture for prediction, it is necessary to encode the future picture used for prediction prior to its coding. Therefore, it is necessary to perform a process of rearranging the pictures in the order of coding.

Here, the method for achieving low delay does not use the B picture in order to eliminate the delay in rearranging the pictures (by processing the future picture first to predict the current picture).

In addition, the processing unit of moving image coding is a block (macro block) of 16 × 16 pixels for a picture. Macroblock types include an intra-macroblock that performs in-picture coding and an inter-macroblock that performs inter-picture coding.

In order to realize low delay in such moving image coding, an intra slice that encodes all the data in the macro block line (slice) in which a certain macro block is arranged horizontally as an intra macro block is used. By shifting the position of the macroblock line to which the intraslice is applied for each picture, the intraslice circulates around the entire picture at regular intervals, so that the entire picture can be refreshed. By such a refresh method, the picture is basically composed of only the P picture, and by not using the I picture, the buffer size can be reduced and the delay due to the buffer can be reduced.

However, at this time, if the macroblocks of slices other than the intraslice of the P picture perform motion compensation using vectors without limitation, even if the intraslice circulates the entire picture, the macroblock after refreshing by the intraslice is intra. By referring to the image at the position before refreshing, if an error occurs at the reference destination, the error will propagate in the spatial direction, and in the worst case, the error may continue to remain on the screen. is there. In order to solve this problem, in Patent Document 2, an error is propagated by prohibiting motion compensation using a non-zero motion vector for a predetermined time for a slice encoded by an intra slice. The method of not disclosing is disclosed.<patcit num="1"><text>Japanese Unexamined Patent Publication No. 60-162392</text></patcit><patcit num="2"><text>Special Fair 06-101841 Gazette</text></patcit>

<p> When applying the method of intra-slicing and motion vector restriction of Patent Document 1 and Patent Document 2 explained in the background degree technique to the international standard ITU-T H.264 (hereinafter referred to as H.264) of moving image coding, Due to a problem peculiar to the H.264 coding method, there was a problem that the error could not be prevented from propagating in the spatial direction. The differences between the MPEG-2 and H.264 coding methods and the problems unique to H.264 will be described below.</p><p> In MPEG-2, when generating an intra slice, there is a syntax called IntraSlice on the slice header, so by turning this flag on, all the macroblocks that make up the specified slice are made up of intra macroblocks. Can be specified. However, in H.264, this syntax does not exist, so it is not possible to generate intraslices. this<u style="single">Intra slice problem</u>I will call it.</p><p> Next, the MPEG-2 intra-macroblock generation method divides the macroblock into 8x8 subblocks and performs DCT (discrete cosine transform) on the pixels contained therein, whereas H.264 In the intra-macroblock generation method of, the compression rate is further improved by adopting a method of using pixel correlation between blocks by in-screen predictive coding in a plurality of modes. That is, even in the coding of the intra macroblock, a predicted image is created from the peripheral pixels of the block to be coded, and the prediction error, which is the difference from the predicted image, is orthogonally converted. Here, when the intra slice is used in MPEG-2, the slice is divided above and below the intra macroblock line, but the intra prediction efficiency is irrespective of the slice division because the intra coding does not use the prediction. It becomes constant. However, in H.264, it is possible to perform intra-coding using prediction from peripheral pixels, and if slice division is performed above and below the intra-macroblock line, peripheral pixels that straddle slices according to the standard. Since the creation of the prediction image from the above is prohibited, the prediction mode in which the prediction from the peripheral pixels is performed, for example, from above is restricted. In this case, the intra-prediction efficiency is reduced by slicing. this<u style="single">Intra-prediction efficiency problem</u>I will call it.</p><p> Further, in H.264, it is stipulated that a deblocking filter is applied between the boundary between the macroblock being encoded and the peripheral macroblock boundary for the purpose of reducing block distortion. Here, the area above the intra macroblock is the refreshed area, and the area below is the deteriorated area where the refresh has not been completed. If a deblocking filter is applied at the boundary between the macroblock and the intra macroblock in this deteriorated area, the pixel component from the deteriorated area flows into the refreshed area, which causes a problem that the refreshed area is deteriorated. this<u style="single">Deblocking filter problem</u>I will call it.</p><p> Further, in MPEG-2, when the forward motion prediction is performed from the past picture in the P picture, the vector restriction that the motion vector is generated only from the refresh area in the past picture is performed. Here, in H.264, a method of generating a motion vector of a color difference component from a motion vector of a luminance component is defined. These depend on the parity of the source and destination fields of the motion vector, respectively, when both fields have the same parity, the former is the top field and the latter is the bottom field, and the former is the bottom field and the latter is the top. In the case of fields, the calculation method is different for the three types. Here, the calculation formula when the reference source is the bottom field and the reference destination is the top field is shown below. MVCy = MVy / 2 + 1/4 However, MVCy is a color difference vector Y direction component, and MVy is a luminance vector Y direction component. Here, when MVy = 0, MVCy = 1/4, and even though the luminance vector is a zero vector, the color difference vector becomes a positive vector, and the reference is made below the same position. Originally, the position of the zero vector in the luminance should be allowed to be referenced because it is within the refresh area, but as described above, the color difference refers to the lower part (deteriorated area), so the pixel component from the deteriorated area is There is a problem that the refresh area deteriorates because it flows into the refresh area. this<u style="single">Vector problem</u>I will call it.</p><p> An object of the present invention is to provide a video coding apparatus that solves the above problems and realizes low delay coding using the moving image coding method H.264.</p>

<p> The present invention employs the following configuration in order to solve the above-mentioned problems. That is, in the moving image coding apparatus using the field coding mode, the present invention forcibly performs intra coding with the forced intra block line determining means for determining the forced intra block line position and the forced intra block line position. The intra-selection means to select, the vector-restricted blockline determining means to determine the vector-restricted blockline position at any blockline in the current field, and the vector-restricted blockline position force the intercoding reference field and reference area. The vector limiting means that limits to, the slice dividing means that forcibly divides the slice with the block line immediately below the block line for which the intra coding is selected in the intra selection means as the boundary, and the slice dividing means that performs the slice division. It is provided with a deblocking filter prohibition means for forcibly prohibiting the deblocking filter processing performed across the block line.</p><p> In the moving image coding device according to the present invention, when the forced intra block line determining means inputs the parity of the current field and the previous forced intra block line position and the parity is the bottom field, the previous forced intra block line position It is equipped with a means to set the line one below the current forced intra-block line position.</p><p> In the moving image coding apparatus according to the present invention, the vector limiting block line determining means takes the previous forced intra block line position as an input and sets the line at the same position as the previous forced intra block line position as the vector limiting block line position. It has the means to do it.</p><p> In the moving image coding device according to the present invention, the intra selection means is a means for forcibly selecting intra coding for a macro block included in the forced intra block line position according to the forced intra block line determining means. It has.</p><p> In the moving image coding apparatus according to the present invention, the vector limiting means is above the vector limiting block line position by inputting the parity of the current field, the parity of the reference field, the slice division position of the reference field, and the vector limiting block line position. In the block of, if the vector limit that allows reference only above the slice division position of the reference field and the parity of the current field is the bottom field and the parity of the reference field is the top field, the position is the same as the vector limit block line position. It has a means to perform vector restriction that makes the block line of a reference field unreferenceable.</p><p> In the moving image coding apparatus according to the present invention, the slice dividing means inputs the parity of the current field, the forced intra block line position, and the vector limiting block line position, and when the parity of the current field is the top field, the vector limiting block. Slice division is performed with the line position and the block line immediately below it as the boundary, and if the parity of the current field is the bottom field, slice division is performed with the forced intra block line position and the block line one level below as the boundary. It has the means. The present invention may be a program that realizes the above functions.</p>

<p>According to the present invention, it is possible to provide a moving image coding device that realizes low delay when a moving image coding method (H.264) that encodes a moving image signal composed of a plurality of fields is used. it can.</p>

Hereinafter, embodiments of the present invention will be described with reference to FIGS. 1 to 15.

FIG. 1 is a diagram illustrating an overall configuration of a moving image coding device 100 according to an embodiment of the present invention. As shown in FIG. 1, the moving image coding device 100 in the present embodiment has a prediction error signal generation means 101, an orthogonal conversion means 102, a quantization means 103, an entropy coding means 104, an inverse quantization means 105, and an inverse orthogonality. Conversion means 106, decoded image generating means 107, deblocking filter means 108, decoded image storage means 109, intra-predicted image generating means 110, inter-predicted image generating means 111, motion vector calculating means 112, coding control and header generating means 113. And the prediction image selection means 114. The outline of each functional unit will be described below.

The prediction error signal generation means 101 divides the current picture of the input moving image data into blocks of 16 × 16 pixels (pixels) (hereinafter referred to as macro blocks (MB)), and the macro block data (hereinafter, also referred to as block data). ) Is obtained, and the prediction error signal is generated from the macroblock data and the macroblock data of the predicted image picture supplied from the predicted image selection means 114. The prediction error signal generating means 101 passes the generated prediction error signal to the orthogonal transforming means 102.

The orthogonal conversion means 102 performs orthogonal conversion processing on the input prediction error signal. The orthogonal transform means 102 supplies the signal separated into horizontal and vertical frequency components by the orthogonal transform process to the quantization means 103. The quantization means 103 quantizes the output of the orthogonal transform means 102. The quantization means 103 reduces the code amount of the prediction error signal by coding and supplies it to the entropy coding means 104 and the inverse quantization means 105.

The entropy coding means 104 outputs the output from the quantization means 103 by entropy coding (variable length coding). Entropy coding is a method of assigning a variable length code according to the frequency of occurrence of symbols.

The inverse quantization means 105 dequantizes the output of the quantization means 103 and then supplies it to the inverse orthogonal transform means 106. The inverse orthogonal transform means 106 performs an inverse orthogonal transform process on the output of the inverse quantization means 105 and then supplies the output to the decoded image generating means 107. By performing the decoding process by the inverse quantization means 105 and the inverse orthogonal transform means 106, a signal having the same degree as the prediction error signal before coding can be obtained.

The decoded image generation means 107 adds the block data of the picture motion-compensated by the inter-prediction image generation means 111 and the prediction error signal decoded by the inverse quantization means 105 and the inverse orthogonal conversion means 106. , The predicted block data of the current image data picture is reproduced and passed to the deblocking filter means 108.

The deblocking filter means 108 applies a filter for reducing block distortion to the decoded image which is the output of the decoded image generating means 107, and passes the decoded image to the decoded image storage means 109.

The decoded image storage means 109 stores the passed block data as data of a new reference picture, and supplies it to the intra prediction image generation means 110, the inter prediction image generation means 111, and the motion vector calculation means 112.

The intra prediction image generation means 110 generates a prediction image from the already encoded peripheral pixels of the picture.

On the other hand, the inter-prediction image generation means 111 blocks the motion-compensated reference picture by motion-compensating the data of the reference picture obtained from the decoded image storage means 109 with the motion vector provided by the motion vector calculation means 112. Generate data.

The motion vector calculation means 112 obtains a motion vector by using the block data in the current image data picture and the block data of the already encoded reference picture obtained from the decoded image storage unit 109. The motion vector is a value indicating the spatial deviation of the block unit obtained by using the block matching technique for searching the position most similar to the current picture from the reference picture in the block unit. The motion vector calculation means 112 passes the obtained motion vector to the inter-prediction image generation means 111.

The block data output from the intra-prediction image generation means 110 and the inter-prediction image generation means 111 is input to the prediction image selection means 114, and either one of the prediction images can be selected. The selected block data is supplied to the prediction error signal generation means 101.

Further, the coding control and the header generation means 113 are subjected to overall coding control and header generation. The intra prediction image generation means 110 is notified of the presence / absence of slice division, the deblocking filter means 108 is notified of the presence / absence of the deblocking filter, the motion vector calculation means 112 is notified of the reference image restriction, and the like. The result is used to generate H.264 header information. The generated header information is passed to the entropy coding means 104, and is output as a stream together with the image data and the motion vector data.

Figure 2 shows the flow of data between the functional blocks of the present invention. In FIG. 2, coding control and header generation means 201 (corresponding to coding control and header generation means 113 in FIG. 1), prediction image selection means 202 (corresponding to prediction image selection means 114 in FIG. 1), motion vector calculation means. Compared to 203 (corresponding to the motion vector calculation means 112 in FIG. 1), it has the functions required for low delay coding in the moving image coding method H.264 described below.

As shown in FIGS. 2 and 3, the coding control and header generation means 201 in the present embodiment includes the forced intra block line determining means 301, the vector limiting block line determining means 302, the slice dividing means 303, and the deblocking filter prohibiting means 304. To be equipped. Each function will be described in detail below.

As shown in FIGS. 2 and 4, the forced intra block line determining means 301 inputs the parity 401 of the current field and the previous forced intra block line position 402, and outputs the forced intra signal 403 to the predicted image selection means 202. .. Further, the forced intra block line position 404 is output to the slice dividing means 303 and the coding control and header control means 201.

As shown in FIG. 5, when the parity of the current field is the bottom field and the previous forced intra block line position 501 is the N-1th line, the forced intra block line determining means 301 is the N line immediately below it. Is the current forced intrablock line position 502. When the macroblock being encoded is included in the forced intrablock line position 502, the forced intra signal 403 is output to the prediction image selection means 202. Further, the forced intra block line position 404 is output to the slice dividing means 303. However, if the previous forced intra block line position 501 has reached the lower limit of the picture, the first block line of the picture is set to the current forced intra block line position 502.

As shown in FIG. 6, the vector limiting block line determining means 302 inputs the previous forced intra block line position 402 and the parity 401 of the current field, and outputs the vector limiting block line position 602 to the motion vector calculating means 203. Similarly, the vector limiting block line position 602 is output to the slice dividing means 303. As shown in FIG. 7, when the current picture is the top field, the vector limiting block line determining means 302 determines that the forced intra block line position 701 existing in the previous bottom field is the N-1th line of the current picture. Let the N-1 line at the same position be the vector limit block line position 702. When the current picture is the bottom field, if the forced intra block line position 703 existing in the previous bottom field is the N-1th line, the N-1 line at the same position is set as the vector restriction block line position 704. Here, when the macro block being coded in the current picture is above the vector limit block line position, the vector limit block line position 602 is output to the motion vector calculation means 203. Similarly, the vector limiting block line position 602 is output to the slice dividing means 303.

As shown in FIG. 8, the slice division means 303 inputs the parity 401 of the current field, the vector restriction block line position 602, and the forced intra block line position 404, and outputs the slice division position 804 to the intra prediction image generation means 110. Further, the slice division position 804 is output to the deblocking filter prohibition means 304 and the coding control and header control means 201.

As shown in FIG. 9, when the parity of the current field is the top field, the slice dividing means 303 performs slice division with the vector limiting block line position 901 and the block line immediately below it as a boundary. If the parity of the current field is the bottom field, slice division is performed with the forced intra block line position 903 and the block line immediately below it as the boundary, and the slice division positions 902 and 904 are output to the intra prediction image generation means. .. Similarly, the slice division positions 902 and 904 are output to the deblocking filter prohibiting means. However, if the slice division positions 902 and 904 reach the lower limit of the picture, it is not necessary to perform slice division. Here, in the coding control and header generation means 201, information that slice division occurs and the deblocking filter is not applied to the slice boundary is generated as H.264 header information. To be precise, H.264 slice header information is generated, and the macroblock number at the beginning of the slice division position is set in the syntax first # mb # in # slice. Also, set disable # deblocking # filter # idc to 2 (meaning that the deblocking filter is not applied to the slice boundary).

As shown in FIG. 10, the deblocking filter prohibition means 304 receives the slice division position 804 obtained by the slice division means 303 as an input, and outputs a filter prohibition signal 1002 to the deblocking filter means.

As shown in FIG. 11, the deblocking filter prohibiting means 304 outputs the filter prohibiting signal 1002 to the deblocking filter means 108 when the macroblock being encoded includes the slice division boundary 1101.

The predictive image selection means 202 in the present embodiment includes an intra selection means 1200 as shown in FIG. The intra selection means 1200 is provided with the switch control means 1201 and receives the forced intra signal 403 output by the forced intra block line determining means 301 as an input, and forcibly predicts the intra for the macroblock instructed by the forced intra signal 403. Select and receive the output of the intra-predictive image generation means 110.

The motion vector calculation means 203 in the present embodiment includes the vector limiting means 1300 as shown in FIG. The vector limiting means outputs the vector data 1305 to the inter-predicted image generating means 111 with the parity 401 of the current field, the parity 1302 of the reference field, the slice division position 1303 of the reference field, and the vector limiting block line position as input 602. To do. As shown in FIG. 14, the vector limiting means 1300 makes it possible to refer only above the slice division position 1401 of the reference field in the upper vector limiting target block 1403 including the vector limiting block line position 1404. Further, when the parity of the current field is the bottom field and the parity of the reference field is the top field, the block line 1402 of the reference field at the same position as the vector restriction block line position 1404 cannot be referred to. The motion vector calculation means 203 in the present embodiment receives the slice division position 804 from the slice division means 303, and prohibits the prediction mode in which the macroblock below the slice boundary makes a prediction beyond the slice boundary. This is due to the H.264 standard. The deblocking filter means 108 in the present embodiment receives the filter prohibition signal 1002 from the deblocking filter prohibition means 304, and prohibits the filter processing across the slice boundary. FIG. 15 shows a configuration example of the bit stream generated from the above embodiment. The video sequence of FIG. 15 consists of repeating bitstream 1500. In this bitstream 1500, refreshing by intra-slicing is performed only on the bottom field image, and the reference destination of the top field is the area where the bottom field has been refreshed. As a result, the top field is also refreshed. The same effect as the above will be obtained. In addition, this bitstream complies with the H.264 standard. That is, if it is a moving image decoding device compatible with H.264, it is possible to reproduce the stream without adding any special processing or device.

Next, the relationship with the problem to be solved by the invention will be described. Regarding the intra-slice problem, the compulsory intra-block line determining means 301 and the intra-selecting means 1200 in this embodiment can forcibly configure an arbitrary macro block line with intra-macroblocks to generate an intra-slice equivalent. Is. Regarding the intra prediction efficiency problem, by setting the slice boundary below the forced intra block line position by the slice dividing means 303 in this embodiment, the prediction mode is restricted for the intra macroblock included in the forced intra block line. I will not receive it. Regarding the deblocking filter problem, the deblocking filter prohibiting means 304 prohibits the filter processing across the slice boundary to prevent the pixel components from the deteriorated region from flowing into the refresh region. Regarding the vector problem, the vector limiting block line determining means 302 and the vector limiting means 1300 in this embodiment make it impossible to refer to the problematic area when the parity of the current field is the bottom field and the parity of the reference field is the top field. This prevents the pixel components from the deteriorated region from flowing into the refresh region. As described above, it is possible to provide a video coding apparatus that realizes low delay coding using the moving image coding method H.264.

Further, the orthogonal transform means described above can be replaced with one capable of performing equivalent processing such as discrete cosine transform, and the entropy coding means can be replaced with one capable of realizing equivalent functions such as variable length coding and arithmetic coding. It is possible.

<figref num="1">It is a figure which shows the whole structure of the moving image coding apparatus which concerns on this embodiment.</figref><figref num="2">It is a figure which shows the function of this invention and the flow of data which concerns on this Example.</figref><figref num="3">It is a figure which shows the structure of the coding control and the header generation means which concerns on this Example.</figref><figref num="4">It is a figure which shows the structure of the forced intra block line determination means which concerns on this Example.</figref><figref num="5">It is a figure which shows the operation of the forced intra block line determination means which concerns on this Example.</figref><figref num="6">It is a figure which shows the structure of the vector restriction block line determination means which concerns on this Example.</figref><figref num="7">It is a figure which shows the operation of the vector restriction block line determination means which concerns on this Example.</figref><figref num="8">It is a figure which shows the structure of the slice division means which concerns on this Example.</figref><figref num="9">It is a figure which shows the operation of the slice division means which concerns on this Example.</figref><figref num="10">It is a figure which shows the structure of the deblocking filter prohibition means which concerns on this Example.</figref><figref num="11">It is a figure which shows the operation of the deblocking filter prohibition means which concerns on this Example.</figref><figref num="12">It is a figure which shows the structure and operation of the intra selection means of this Example.</figref><figref num="13">It is a figure which shows the structure of the vector limiting means which concerns on this Example.</figref><figref num="14">It is a figure which shows the operation of the vector limiting means which concerns on this Example.</figref><figref num="15">It is a figure which shows the structure of the generated bit stream which concerns on this Example.</figref>

Code description

100 video coding device 101 Forecast error signal generation means 102 Orthogonal conversion means 103 Quantization means 104 Entropy encoding means 105 Inverse quantization means 106 Inverse orthogonal conversion means 107 Decoded image generation means 108 Deblocking filter means 109 Decrypted image storage means 110 Intra-predictive image generation means 111 Inter-predictive image generation means 112 Motion vector calculation means 113 Coding control and header generation means 114 Predictive image selection method 201 Coding control and header generation means 202 Predictive image selection method 203 Motion vector calculation means 301 Forced intra block line determination means 302 Vector limit block line determination means 303 Slice division means 304 Deblocking filter prohibition means 401 Current field parity 402 Last forced intra block line position 403 Forced intra signal 404 Forced intra-block line position 501 Last forced intra block line position 502 Current forced intra block line position 602 Vector limit block line position 701 Last forced intra block line position 702 Top field vector limit block line position 703 Last forced intra block line position 704 Top field vector limit block line position 804 Slice division position 901 Vector limit block line position 902 Top field slice division position 903 Forced intra block line position 904 Bottom field slice division position 1002 Filter prohibition signal 1101 Deblocking filter prohibited position 1200 intra selection method 1201 Switch control means 1300 Vector limiting means 1302 Reference field parity 1303 Reference field slice split position 1305 vector data 1401 Reference field slice split position 1402 Unreferenceable reference field block line 1403 Vector restricted block 1404 Vector limit block line position 1500 bitstream