JP2008271422A - Moving image predictive coding apparatus, method and program, and moving image predictive decoding apparatus, method and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To more accurately approximate a predictive signal to a target signal, by correcting the predictive signal with less auxiliary information. <P>SOLUTION: A coding apparatus is provided with a regional dividing means for dividing an input image into a plurality of regions; a predictive signal generating means for generating an approximate signal for a target pixel signal in a target region to be processed in the plurality of regions; a residual signal generating means for generating a residual signal between the approximate signal and the target pixel signal; and a coding means for coding the residual signal, wherein the predictive signal generating means determines a predictive signal for the target region, uses a target adjacent pixel group, constituted of reproduced pixels adjacent to the target region and a predictive adjacent pixel group, constituted of reproduced pixels in the same positional relation, to a reference region that includes the predictive signal, as the positional relation between the target adjacent pixel group and the target region, to determine a parameter approximating the predictive adjacent pixel group to the target adjacent pixel group, and converts the predictive signal, on the basis of the parameter, generates an approximate signal for the target pixel signal in the target region. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a moving picture predictive coding apparatus, method and program, and a moving picture predictive decoding apparatus, method and program.

  In order to efficiently transmit and store moving image data, a compression encoding technique is used. In the case of moving images, MPEG1, MPEG2, MPEG4, H.264, and so on. 261-H. H.264 is widely used. In encoding of a moving image, a prediction signal for a target image to be encoded is generated using another image adjacent in the time direction, and a difference between the image signal of the target image and the prediction signal is encoded. As a result, a reduction in the amount of data is realized.

  For example, H.C. In H.264, an image of one frame is divided into block areas each having 16 × 16 pixels, and the image is encoded in units of blocks. In interframe coding, motion detection is performed on a target block in an image to be encoded using another frame that has been encoded and restored as a reference image, and a prediction signal with the least error is determined. Next, a difference value between the image signal of the target block and the prediction signal is obtained, and discrete cosine transform and quantization processing are performed. Then, the quantized discrete cosine transform coefficient and the motion vector for specifying the prediction signal are entropy-coded to generate encoded data.

  In order to increase the coding efficiency within the screen, predictive coding is also performed within the screen. H. In the case of H.264, a prediction signal is generated using an already reproduced image signal (reconstructed from image data once compressed) of an adjacent pixel group in the same screen as the target block, and the prediction signal is generated from the target block. The difference (residual signal) obtained by subtraction from the image signal is encoded.

  A specific method for generating a prediction signal between screens is described in, for example, Patent Document 1, and a specific method for generating a prediction signal in a screen is described in, for example, Patent Document 2.

On the other hand, when there is a partial luminance change such as fade-in and fade-out, the luminance of the prediction signal between screens is further corrected, and then the difference between the image signal of the target block and the prediction signal is encoded. Techniques are known. For example, Patent Document 3 discloses a related art relating to such luminance correction.
US Patent Publication No. 52008220 US Pat. No. 6,765,964 US Pat. No. 6,934,331

  However, in the luminance correction method in the prior art, approximate parameters (also referred to as correction parameters and auxiliary information) used for the correction are indispensable to correct the prediction signal correctly on the receiving side. The parameters need to be encoded with the residual signal and stored or transmitted. Therefore, the amount of compressed data increases by the approximate parameter.

  In order to improve the performance of the prediction signal, it is desirable to divide the image into smaller areas than before and perform luminance correction on each small area. However, since the data amount of the approximate parameter increases rapidly in proportion to the increasing number of small regions, the entire data amount increases.

  Furthermore, in the prior art, for example, when the offset value is added after scaling the luminance signal, two parameters are included as approximate parameters. If the image signal of the target block changes non-linearly with respect to the prediction signal, In the case shown, the number of parameters of the approximate parameter is further increased, and the amount of auxiliary information such as the approximate parameter is further increased.

  The present invention solves the above-described problem, corrects a prediction signal with a small amount of auxiliary information, and can approximate a target signal with higher accuracy, a video prediction encoding apparatus, a method and a program, and a video prediction decoding apparatus It is an object to provide a method and a program.

  The moving picture predictive coding apparatus according to the present invention includes a region dividing unit that divides an input image into a plurality of regions, and a target pixel signal of a target region that is a processing target among the plurality of regions divided by the region dividing unit. Generated by the prediction signal generation means, the residual signal generation means for generating the residual signal between the approximate signal generated by the prediction signal generation means and the target pixel signal, and the residual signal generation means Encoding means for encoding the residual signal, and the prediction signal generation means determines a prediction signal for the target area by a predetermined method, and a target adjacent pixel group composed of already reproduced pixels adjacent to the target area; The predicted adjacent pixel group is defined as the target adjacent pixel group using the predicted adjacent pixel group including the already-reproduced pixels, which is in the same positional relationship as the positional relationship between the target region and the target adjacent pixel group with respect to the reference region including the predicted signal. In Determining the approximate parameters for causing the similar, by converting a prediction signal for the target region based on the approximation parameters, and generates an approximate signal for the target pixel signals of the target area, characterized in that.

  The moving picture predictive encoding apparatus according to the present invention includes an area dividing unit that divides an input image into a plurality of areas, and a target pixel of a target area that is a processing target among the plurality of areas divided by the area dividing unit. Prediction signal generation means for generating an approximate signal for the signal, residual signal generation means for generating a residual signal between the approximate signal generated by the prediction signal generation means and the target pixel signal, and generation by the residual signal generation means Encoding means for encoding the residual signal, and the prediction signal generating means uses a region composed of pixels reproduced before the target region as a reference region, and has already been reproduced pixels adjacent to the target region A pixel group having the highest correlation among the reference regions is derived as a predicted adjacent pixel group, and is in the same positional relationship as the positional relationship between the target adjacent pixel group and the target region. Is obtained as a prediction signal for the target region, and approximate parameters for approximating the predicted adjacent pixel group to the target adjacent pixel group using the predicted adjacent pixel group and the target adjacent pixel group are obtained. An approximate signal for the target pixel signal of the target area is generated by determining and converting a prediction signal for the target area based on the approximate parameter.

  The video predictive decoding apparatus according to the present invention includes a data analysis unit that extracts encoded data of a residual signal for a target region to be processed from compressed data, and a residual signal extracted by the data analysis unit. A residual signal restoring means for restoring a reproduction residual signal from the encoded data of the target area, a prediction signal generating means for generating an approximate signal for the target pixel signal in the target area, an approximate signal generated by the prediction signal generating means and the residual signal Image restoration means for restoring the target pixel signal of the target area by adding the reproduction residual signal restored by the difference signal restoration means, and the prediction signal generating means The prediction signal is obtained, and the target adjacent pixel group including the already reproduced pixels adjacent to the target region and the reference region including the prediction signal are the same as the positional relationship between the target region and the target adjacent pixel group. An approximate parameter for approximating the predicted adjacent pixel group to the target adjacent pixel group is determined using the predicted adjacent pixel group composed of already reproduced pixels, and a prediction signal for the target region based on the approximate parameter To generate an approximate signal for the target pixel signal in the target region.

  The moving picture predictive decoding apparatus according to the present invention includes a data analysis unit that extracts encoded data of a residual signal for a target region that is a processing target from compressed data, and a residual that is extracted by the data analysis unit. Residual signal restoring means for restoring the reproduction residual signal from the encoded data of the difference signal, predicted signal generating means for generating an approximate signal for the target pixel signal in the target region, and the approximate signal generated by the predicted signal generating means And a restoration residual signal restored by the residual signal restoration means, and an image restoration means for restoring the target pixel signal of the target area, and the prediction signal generation means reproduces before the target area. The region composed of the processed pixels is used as a reference region, and the pixel group having the highest correlation from the reference region is predicted adjacent to the target adjacent pixel group composed of already reproduced pixels adjacent to the target region. Deriving as a pixel group, determining an approximate parameter for approximating the predicted adjacent pixel group to the target adjacent pixel group using the predicted adjacent pixel group and the target adjacent pixel group, and the positional relationship between the target adjacent pixel group and the target region An approximate signal for the target region is generated by converting pixel signals in the reference region corresponding to the predicted adjacent pixel group having the same positional relationship based on the approximate parameter.

  The moving picture predictive coding method according to the present invention is a moving picture predictive coding method executed by a moving picture predictive coding apparatus, and includes a region dividing step for dividing an input image into a plurality of regions, and a region dividing step. A prediction signal generation step for generating an approximate signal for a target pixel signal in a target region that is a processing target of the plurality of divided regions, and a residual between the approximate signal generated by the prediction signal generation step and the target pixel signal A residual signal generation step for generating a signal, and an encoding step for encoding the residual signal generated by the residual signal generation step. In the prediction signal generation step, the prediction of the target region is performed by a predetermined method. The position relationship between the target area and the target adjacent pixel group is determined with respect to the target adjacent pixel group including the already reproduced pixels adjacent to the target area and the reference area including the prediction signal The approximate parameter for approximating the predicted adjacent pixel group to the target adjacent pixel group is determined using the predicted adjacent pixel group consisting of the already reproduced pixels that are in the same positional relationship as the target region based on the approximate parameter. By converting the predicted signal, an approximate signal for the target pixel signal in the target region is generated.

  The moving picture predictive coding method according to the present invention is a moving picture predictive coding method executed by a moving picture predictive coding apparatus, and includes a region dividing step for dividing an input image into a plurality of regions, A prediction signal generation step for generating an approximate signal for a target pixel signal of a target region that is a processing target among a plurality of regions divided by the step, and the approximate signal generated by the prediction signal generation step and the target pixel signal A residual signal generation step for generating a residual signal; and an encoding step for encoding the residual signal generated by the residual signal generation step. In the prediction signal generation step, the residual signal is reproduced before the target region. A pixel that has the highest correlation among the reference area with respect to the target adjacent pixel group consisting of the already reproduced pixels adjacent to the target area. Is obtained as a predicted adjacent pixel group, and the pixel signal of the reference region corresponding to the predicted adjacent pixel group that is in the same positional relationship as the positional relationship between the target adjacent pixel group and the target region is acquired as a predicted signal for the target region, By determining the approximate parameter for approximating the predicted adjacent pixel group to the target adjacent pixel group using the predicted adjacent pixel group and the target adjacent pixel group, and converting the prediction signal for the target region based on the approximate parameter An approximate signal for the target pixel signal in the target region is generated.

  A moving picture predictive decoding method according to the present invention is a moving picture predictive decoding method executed by a moving picture predictive decoding apparatus, and includes encoded data of a residual signal for a target region to be processed from compressed data. A data analysis step for extracting the residual signal, a residual signal restoration step for restoring the reproduction residual signal from the encoded data of the residual signal extracted by the data analysis step, and generating an approximate signal for the target pixel signal in the target region A prediction signal generation step; and an image restoration step of restoring the target pixel signal of the target region by adding the approximate signal generated by the prediction signal generation step and the reproduction residual signal restored by the residual signal restoration step; In the prediction signal generation step, the prediction signal for the target area is obtained by a predetermined method, and the reproduced signal adjacent to the target area is obtained. Predicted adjacent pixels using a predicted adjacent pixel group composed of already-reproduced pixels that are in the same positional relationship as the positional relationship between the target region and the target adjacent pixel group with respect to the target adjacent pixel group and the reference region including the prediction signal. Determining an approximate parameter for approximating the group to the target adjacent pixel group, and generating an approximate signal for the target pixel signal of the target region by converting a prediction signal for the target region based on the approximate parameter; It is characterized by that.

  The moving picture predictive decoding method according to the present invention is a moving picture predictive decoding method executed by a moving picture predictive decoding apparatus, and includes coding of a residual signal for a target region to be processed from compressed data. A data analysis step for extracting the encoded data, a residual signal restoration step for restoring the reproduction residual signal from the encoded data of the residual signal extracted by the data analysis step, and an approximation signal for the target pixel signal of the target region Image restoration for restoring the target pixel signal of the target region by adding the prediction signal generation step to be generated, and the approximate signal generated by the prediction signal generation step and the reproduction residual signal restored by the residual signal restoration step A prediction signal generating step, wherein an area composed of pixels reproduced before the target area is set as a reference area, and the target area A pixel group having the highest correlation from the reference region is derived as a predicted adjacent pixel group with respect to a target adjacent pixel group including adjacent already-reproduced pixels, and predicted adjacent using the predicted adjacent pixel group and the target adjacent pixel group Approximate parameter for approximating the pixel group to the target adjacent pixel group is determined, and the pixel signal of the reference area corresponding to the predicted adjacent pixel group that is in the same positional relationship as that of the target adjacent pixel group and the target area is an approximate parameter An approximate signal for the target region is generated by performing conversion based on the above.

  A moving image predictive coding program according to the present invention is a moving image predictive coding program for causing a computer to function as the moving image predictive coding device according to claim 1 or 2.

  The moving picture predictive decoding program according to the present invention is a moving picture predictive decoding program for causing a computer to function as the moving picture predictive decoding device according to claim 3 or 4.

  According to the present invention, the target adjacent pixel group composed of the already reproduced pixels adjacent to the target area and the reference area including the prediction signal are in the same positional relationship as the positional relation between the target area and the target adjacent pixel group. An approximate parameter for approximating the predicted adjacent pixel group to the target adjacent pixel group is determined using the predicted adjacent pixel group composed of pixels, and an approximation for the target region is performed by converting the predicted signal using the approximate parameter Find the signal. For this reason, it is possible to correctly derive the approximate parameter even on the receiving side, and it is not necessary to store or transmit the approximate parameter together with the residual signal, and it is possible to suppress an increase in the data amount due to the approximate parameter.

  Along with this, even when the image is divided into smaller areas than before and luminance correction is performed for each small area, or even when the target signal shows a nonlinear change with respect to the prediction signal, the approximate parameter increases. It is possible to generate a high-precision approximate signal while avoiding an increase in the amount of auxiliary information data caused.

  As described above, according to the present invention, a highly accurate approximate signal can be generated with a small amount of auxiliary information. For this reason, the residual signal is reduced, and the code amount of the entire moving image can be greatly reduced.

  Embodiments according to the present invention will be described below with reference to FIGS.

[About video predictive coding apparatus]
FIG. 1 shows a block diagram of a video predictive coding apparatus 100 according to the present embodiment. As shown in FIG. 1, the moving picture predictive coding apparatus 100 includes an input terminal 101, a block divider 102, a signal generator 103, a frame memory 104, a subtractor 105, a converter 106, a quantizer 107, and an inverse quantization. A counter 108, an inverse transformer 109, an adder 110, an entropy encoder 111, and an output terminal 112.

[Outline of operation in video predictive coding apparatus]
Hereinafter, the operation of each component of the moving picture predictive coding apparatus 100 will be described. A moving image signal composed of a plurality of images is input to the input terminal 101. With respect to the input image, the block divider 102 divides the image to be encoded into a plurality of regions. In this embodiment, for example, the block is divided into 8 × 8 pixels, but may be divided into other block sizes or shapes. Next, the signal generator 103 generates an approximate signal for a region to be encoded (hereinafter referred to as “target block”). In the present embodiment, the signal generator 103 generates an approximate signal at least one of inter-screen prediction and intra-screen prediction. The generation process will be described later.

  The approximate signal generated by the signal generator 103 is sent to the subtractor 105. The subtractor 105 generates a residual signal by subtracting the approximate signal input via the line L103 from the signal of the target block input via the line L102. This residual signal is subjected to discrete cosine transform by the transformer 106, and each coefficient after the transformation is quantized by the quantizer 107. Each quantized coefficient after conversion is encoded by the entropy encoder 111 and transmitted from the output terminal 112.

  On the other hand, in order to perform at least one of inter-frame prediction and intra-screen prediction for the subsequent target block, each transformed coefficient quantized by the quantizer 107 is subjected to inverse transformation processing and restored as follows. That is, each quantized coefficient after transform is inversely quantized by the inverse quantizer 108 and then inverse discrete cosine transformed by the inverse transformer 109. Thereby, the residual signal is restored. The restored residual signal is added by the adder 110 to the prediction signal input from the line L103. As a result, the signal of the target block is reproduced, and the reproduced signal is stored in the frame memory 104. In this embodiment, the converter 106 and the inverse converter 109 are used. However, other conversion processes in place of these converters may be used. In some cases, the moving image predictive coding apparatus 100 may not include the converter 106 and the inverse transformer 109 and may adopt a mode in which the conversion process and the inverse conversion process are not performed.

  Next, the configuration and operation of the signal generator 103 provided in the video predictive encoding device 100 will be described with reference to FIG. As shown in FIG. 2, the signal generator 103 includes a prediction signal generator 201 that generates a prediction signal, a parameter estimator 202 that estimates an approximate parameter, and an approximate signal generator 203 that generates an approximate signal. Hereinafter, operations of the signal generator 103 will be described in the order of (1) inter-screen prediction and (2) intra-screen prediction.

[(1) Inter-screen prediction]
In the inter-screen prediction of the present embodiment, the prediction signal generator 201 provided in the signal generator 103 uses a reconstructed image that has been encoded in the past as a reference image, and uses the reference image to calculate the target block. Motion information that gives a prediction signal with the smallest error (also referred to as “motion vector”, but “motion information” is used in this specification) is obtained. Specifically, the image signal of the target block is input to the prediction signal generator 201 via the line L102, and the image signal of the reference image is input to the prediction signal generator 201 via L104, and the prediction signal generator 201 extracts the target block from the reference image. The motion information that gives the prediction signal with the smallest error (the optimal prediction signal) is detected by the block matching method. An example of motion information detection is described in Patent Document 1 described above. In this embodiment, one or a plurality of images restored after being encoded in the past is used as a reference image. However, depending on the case, the target block is subdivided, and the subdivided subregions are Prediction may be performed to obtain motion information. In this case, in each of the various division methods, when the target block is subdivided and inter-frame prediction is performed to obtain motion information, the total error of the entire target block is minimized (the most efficient for the entire target block). (Good) It is desirable to determine the division method and the motion information obtained by the division method.

  The motion information determined in this way is sent to the entropy encoder 111 in FIG. 1 via the line L112, encoded by the entropy encoder 111, and sent from the output terminal 112. The motion information is sent to the parameter estimator 202 via the line L201. The parameter estimator 202 estimates an approximate parameter for further approximating the optimal prediction signal given by the motion information determined by the prediction signal generator 201 as follows.

  The operation of the parameter estimator 202 will be described with reference to FIG. An area 302 illustrated in FIG. 3 is a target block 302 that is a current processing target. Of the target image 301, a region 304 that does not include the target block 302 is a region that includes pixels that have already been encoded and restored (hereinafter referred to as “reproduced pixels”), and a region 305 that includes the target block 302 is encoded. And an area composed of pixels that have not yet been restored.

  The parameter estimator 202 acquires the pixel group 303 adjacent to the target block 302 from the frame memory 104 via the line L104 based on the position information of the target block 302. In the present embodiment, the pixel group 303 is referred to as a “target adjacent pixel group 303”, and is composed of already-reproduced pixels in an inverted “L” -shaped region including N / 2 pixels adjacent to the top and left of the target block 302. ing. Here, N is the number of vertical or horizontal pixels of the target block 302.

  Next, the parameter estimator 202 acquires a pixel group 308 adjacent to the optimal prediction signal 307 given by the motion information in the reference image 306 based on the motion information input via the line L201. In the present embodiment, the pixel at the upper left corner of the prediction signal 307 is specified by the motion information, and an inverted “L” pixel group 308 that is N / 2 pixels away from the upper left pixel upward and left is acquired. This pixel group 308 is referred to as a “predicted adjacent pixel group 308”. As can be seen from FIG. 3, the positional relationship between the prediction signal 307 and the predicted adjacent pixel group 308 is equal to the positional relationship between the target block 302 and the target adjacent pixel group 303. In addition to the inverted “L” shape, a block composed of N × N pixels positioned above the target block 302 in the target image 301 may be set as the target adjacent pixel group, or positioned to the left of the target block 302. A block composed of N × N pixels may be used as a target adjacent pixel group, and further, a block composed of N × N pixels located on the target block 302 and N × located on the left of the target block 302. A total of two blocks of N pixels may be used as the target adjacent pixel group. In these cases, the predicted adjacent pixel group 308 needs to correspond to the target adjacent pixel group 303.

ここで、ｘ（ｉ，ｊ）は対象隣接画素群３０３にある（ｉ，ｊ）番目の画素であり、ｙ（ｉ，ｊ）は予測隣接画素群３０８にある（ｉ，ｊ）番目の画素である。また、近似パラメータｓとｖを用いて予測隣接画素群３０８を変換する。これら近似パラメータｓとｖは最小２乗法で決定する。即ち、以下の式（２）に示す評価式の最小点を与えるｓとｖを求める。ｓとｖはそれぞれ式（３）と式（４）に示すように対象隣接領域と予測隣接領域の全ての画素によって決定される。ここで、Ｍは対象隣接領域（または予測隣接領域）に含まれる画素の数である。

なお、式（１）の代わりにそれ以外の近似式（例えば上記の式（５））を用いてもよい。式（５）を用いる場合、近似パラメータはａ，ｂ，ｃとなる。この場合、予測信号を湾曲させて近似することができる。 Further, the parameter estimator 202 uses the acquired target adjacent pixel group 303 and predicted adjacent pixel group 308 to derive approximate parameters as follows. In the present embodiment, the predicted adjacent pixel group 308 is approximated to the target adjacent pixel group 303 by the following equation (1).

Here, x (i, j) is the (i, j) th pixel in the target adjacent pixel group 303, and y (i, j) is the (i, j) th pixel in the predicted adjacent pixel group 308. It is. Also, the predicted adjacent pixel group 308 is converted using the approximate parameters s and v. These approximate parameters s and v are determined by the least square method. That is, s and v giving the minimum point of the evaluation formula shown in the following formula (2) are obtained. s and v are determined by all the pixels in the target adjacent region and the predicted adjacent region as shown in Equation (3) and Equation (4), respectively. Here, M is the number of pixels included in the target adjacent region (or predicted adjacent region).

Note that an approximate expression other than that (for example, the above expression (5)) may be used instead of the expression (1). When equation (5) is used, the approximate parameters are a, b, and c. In this case, the prediction signal can be approximated by curving.

なお、近似信号生成器２０３は、必ずしもフレームメモリ１０４から予測信号を取得する必要はなく、例えば予測信号生成器２０１により決定された最適な予測信号を予測信号生成器２０１から取得してもよい。本実施形態では、上記の近似パラメータｓとｖは、符号化されて蓄積又は送信されることはない。 The approximate parameter obtained in this way is sent to the approximate signal generator 203 via the line L202 in FIG. 2 together with the motion information. The approximate signal generator 203 generates an approximate signal using the approximate parameters. Specifically, the approximate signal generator 203 acquires the prediction signal 307 of FIG. 3 from the frame memory 104 via the line L104 based on the motion information sent via the line L202. Each pixel of the prediction signal 307 is converted into y ^* (i, j) in the following equation (6), and the prediction signal is converted as shown in equation (6) using the approximation parameters s and v. Pixel x ^* (i, j) is obtained. The approximate signal obtained in this way is sent to the subtracter 105 in FIG. 1 via the line L103.

Note that the approximate signal generator 203 does not necessarily need to acquire a prediction signal from the frame memory 104. For example, an optimal prediction signal determined by the prediction signal generator 201 may be acquired from the prediction signal generator 201. In the present embodiment, the approximate parameters s and v are not encoded and stored or transmitted.

  FIG. 4 shows a flowchart of inter-screen prediction processing for generating an approximate signal, which is executed by the signal generator 103. In step 411 of FIG. 4, the prediction signal generator 201 obtains an optimal prediction signal for the target block by the block matching method as described above. Thereby, motion information indicating the address of the optimal prediction signal is determined and input to the parameter estimator 202. In step 412, the parameter estimator 202 obtains pixels of the target adjacent pixel group adjacent to the target block. As the target adjacent pixel group, for example, the target adjacent pixel group 303 shown in FIG. 3 is used. In step 413, the parameter estimator 202 acquires the prediction adjacent pixel group 308 of FIG. 3 adjacent to the prediction signal specified by the motion information determined in step 411.

  Next, the parameter estimator 202 obtains an approximate parameter using the target adjacent pixel group and the predicted adjacent pixel group (step 414). More specifically, approximate parameters s and v are obtained using equations (3) and (4) that give the minimum value of equation (2). The obtained approximate parameters s and v are input to the approximate signal generator 203 together with the motion information.

In step 415, the approximate signal generator 203 uses the approximate parameters s and v to convert the prediction signal 307 of FIG. 3 specified by the motion information as shown in Expression (6), and approximate signal x ^* ( i, j) is calculated. Finally, the approximate signal generator 203 sends the calculated approximate signal to the subtractor 105 (step 416). The above processing of FIG. 4 may be executed for all the target blocks in the encoding target image, or may be executed for a part of the target blocks in the encoding target image.

[(2) In-screen prediction]
Next, the case of intra prediction according to the present embodiment will be described. In the intra prediction, the prediction signal generator 201 provided in the signal generator 103 generates an intra prediction signal using already reproduced pixel values spatially adjacent to the target block. The prediction signal generator 201 acquires the already-reproduced pixel signal in the same screen as the target block from the frame memory 104, and uses the nine methods shown in FIGS. 5 (A) to 5 (I) for example. The prediction signal is generated, the prediction signal closest to the target block (with a small error) is determined from the generated nine prediction signals, and the closest prediction signal (optimal prediction signal) is generated. The interpolation method is determined from the above nine methods.

  FIGS. 5A to 5I are schematic diagrams showing a specific method for generating a prediction signal by intra prediction, and the specific method is described in, for example, Patent Document 2 described above. ing. 5A to 5I, for example, in FIG. 5A, a block 502 indicates a target block, and a pixel group 501 including pixels A to M adjacent to the boundary of the target block 502 is It is an image signal that has already been reproduced in the past processing (an already reproduced image signal). In FIG. 5A, all the image signals of the four pixels in the leftmost column in the target block 502 are all directly above the pixel A, as indicated by a downward arrow in each column of the target block 502. It is the same as the image signal. Similarly, the other columns in the target block 502 are assumed to be the same as the image signals of the pixels B to D immediately above the respective columns. In this way, a prediction signal for the target block 502 is generated. In FIG. 5B, all the image signals of the four pixels in the uppermost row in the target block 504 are placed to the left of the row, as indicated by a right-pointing arrow in each column of the target block 504. The same as the image signal of a certain pixel I. Similarly, the other rows in the target block 504 are the same as the image signals of the pixels J to L on the left side. In this way, a prediction signal for the target block 504 is generated. Further, in FIG. 5C, the image signal of each pixel of the target block 506 is the average value of the image signals of a total of eight pixels A to D and I to L, so that the prediction signal for the target block 506 is obtained. Is generated. In other FIGS. 5 (D) to 5 (I), a predetermined weighted average arithmetic expression correlated with the arrows is used for the already reproduced image signals of a plurality of pixels located in the directions along the arrows in each drawing. By performing a weighted average calculation, an image signal of each pixel of the target block is obtained, and a prediction signal for the target block is generated.

  Thus, for each of the nine prediction signals generated by the method shown in FIGS. 5A to 5I, the difference from the pixel signal of the target block is calculated, and the prediction signal having the smallest difference is calculated. The optimum prediction signal is determined, and the method used for obtaining the optimum prediction signal is determined as the optimum interpolation method.

  Information regarding the interpolation method determined in this way is sent to the entropy encoder 111 in FIG. 1 via the line L112, encoded by the entropy encoder 111, and then output from the output terminal 112. Further, the information on the interpolation method is sent to the parameter estimator 202 via the line L201, and the parameter estimator 202 further approximates the optimum prediction signal determined by the prediction signal generator 201 as follows. Estimate approximate parameters for

  The operation of the parameter estimator 202 will be described with reference to FIG. A block 601 is a target block that is a current processing target. The parameter estimator 202 acquires the pixel group 602 adjacent to the target block 601 from the frame memory 104 via the line L104 based on the position information of the target block 601. In the present embodiment, the pixel group 602 adjacent to the target block 601 is referred to as a “target adjacent pixel group 602”, and the target adjacent pixel group 602 includes an N / 2 pixel adjacent to the top and left of the target block 601. It consists of already reproduced pixels in the “L” -shaped region. Here, N is the number of vertical or horizontal pixels of the target block 601.

  Next, the parameter estimator 202 uses the pixels 603 to 621 adjacent to the target adjacent pixel group 602 in the interpolation method specified by the information regarding the interpolation method input via the line L201, and sets the target adjacent pixel group 602. A corresponding interpolation signal is generated. The generated interpolation signal has the same shape as the target adjacent pixel group 602 and is composed of the same number of pixels. The generated interpolation signal is a “predicted adjacent pixel group” in the case of intra prediction. In addition to the above inverted “L” character, a block composed of N × N pixels positioned on the target block 601 may be used as a target adjacent pixel group, or N × positioned on the left of the target block 601. A block composed of N pixels may be set as a target adjacent pixel group, and further, a block composed of N × N pixels positioned on the target block 601 and N × N pixels positioned on the left of the target block 601. A total of two blocks of blocks composed of the pixels may be set as the target adjacent pixel group. In these cases, the predicted adjacent pixel group needs to correspond to the target adjacent pixel group.

  Next, the parameter estimator 202 derives approximate parameters using the target adjacent pixel group 602 and the predicted adjacent pixel group generated by interpolation as described above as follows. In the present embodiment, the predicted adjacent pixel group is approximated to the target adjacent pixel group 602 by the above-described equation (1).

  In equation (1), x (i, j) is the (i, j) th pixel in the target adjacent pixel group 602, and y (i, j) is the (i, j) th in the predicted adjacent pixel group. Pixels. s and v are approximate parameters, and these approximate parameters s and v are obtained by calculating s and v that give the minimum point of the evaluation formula shown in the above-described formula (2). Instead of the formula (1), another approximate formula (for example, the formula (5) described above) may be used. When equation (5) is used, the approximate parameters are a, b, and c. In this case, the prediction signal can be approximated by curving.

The approximate parameter obtained in this way is sent to the approximate signal generator 203 via the line L202 in FIG. 2 together with information on the interpolation method. The approximate signal generator 203 generates an approximate signal using the approximate parameters. Specifically, the approximate signal generator 203 obtains pixels (pixels A to M in FIG. 5) adjacent to the target block from the frame memory 104 via the line L104, and relates to an interpolation method input via the line L202. A prediction signal is generated based on the information. Each pixel of this prediction signal is converted into y ^* (i, j) in the above-described equation (6), and the prediction signal is converted as shown in equation (6) using the approximation parameters s and v. x ^* (i, j) is obtained. The approximate signal obtained in this way is sent to the subtracter 105 in FIG. 1 via the line L103.

  Note that the approximate signal generator 203 is not necessarily required to generate a prediction signal. For example, an optimal prediction signal determined by the prediction signal generator 201 may be acquired from the prediction signal generator 201. In the present embodiment, the approximate parameters s and v are not encoded and stored or transmitted.

  FIG. 7 shows a flowchart of the intra-screen prediction process for generating an approximate signal, which is executed by the signal generator 103. In step 711 of FIG. 7, the prediction signal generator 201 uses the neighboring pixels to select the optimum for the target block from among the nine methods shown in FIGS. 5A to 5I as described above. A prediction signal is determined, and an interpolation method (that is, an optimal interpolation method) capable of obtaining the optimal prediction signal is determined. Thereby, an optimal interpolation method is determined, and information regarding the interpolation method is input to the parameter estimator 202. In step 712, the parameter estimator 202 obtains pixels of the target adjacent pixel group adjacent to the target block. As the target adjacent pixel group, for example, a target adjacent pixel group 602 shown in FIG. 6 is used. In step 713, the parameter estimator 202 uses the neighboring pixels (pixels 603 to 621 in FIG. 6) of the target adjacent pixel group according to the interpolation method determined in step 711, and predictive adjacent corresponding to the target adjacent pixel group 602. An interpolation signal for a pixel group is generated.

  Next, the parameter estimator 202 obtains an approximate parameter using the target adjacent pixel group and the predicted adjacent pixel group (step 714). Specifically, the approximation parameters s and v are obtained using Equation (3) and Equation (4) that give the minimum value of Equation (2) described above. The obtained approximate parameters s and v are input to the approximate signal generator 203 together with information on the interpolation method determined in step 711.

In step 715, the approximate signal generator 203 generates an approximate signal using the approximate parameter. Specifically, the approximate signal generator 203 obtains pixels (pixels A to M in FIG. 5) adjacent to the target block from the frame memory 104 via the line L104, and relates to an interpolation method input via the line L202. A prediction signal is generated based on the information. Further, the approximate signal generator 203 uses the approximate parameters s and v to convert the predicted signal according to the above-described equation (6), and calculates an approximate signal x ^* (i, j). Finally, the approximate signal generator 203 sends the calculated approximate signal to the subtractor 105 (step 716). The above processing of FIG. 7 may be executed for all the target blocks in the encoding target image, or may be executed for a part of the target blocks in the encoding target image.

[Another form of signal generator]
FIG. 8 is a block diagram showing another form of the signal generator 103 according to the present embodiment. The signal generator 103 shown in FIG. 8 is not connected to the line L102 and the line L112 shown in FIG. For other configurations, the signal generator 103 of FIG. 8 is the same as the signal generator of FIG. Note that the moving picture predictive encoding apparatus may be configured to include both the signal generator of FIG. 8 and the signal generator of FIG. 2 and switch these signal generators to operate. Hereinafter, the operation of the signal generator 103 shown in FIG. 8 will be described.

  A prediction signal generator 801 provided in the signal generator 103 generates an optimal prediction signal for the target block. The process of the prediction signal generator 801 will be described using FIG. 3 again. The target block 302 that is the current processing target is specified by position information managed by a control unit (not shown) of the entire system. The prediction signal generator 801 acquires a pixel group 303 adjacent to the target block 302 from the frame memory 104 via the line L104 based on the position information of the target block 302. In the present embodiment, the pixel group 303 is referred to as a “target adjacent pixel group 303”, and this target adjacent pixel group is an inverted “L” -shaped region including N / 2 pixels adjacent to the top and left of the target block 302. It is composed of some already reproduced pixels. Here, N is the number of vertical or horizontal pixels of the target block 302.

  Next, the prediction signal generator 801 obtains a pixel group having the smallest error with respect to the target adjacent pixel group 303 within the range of the reference region 309 in the reference image 306 different from the target image 301. Specifically, the prediction signal generator 801 performs matching between the pixel group and the target adjacent pixel group 303 while shifting the pixel group within the range of the reference region 309 by one pixel, and the difference between the pixel groups. A pixel group having the smallest absolute value sum is found, and a pixel group having the smallest sum of absolute values of the differences is used as a prediction signal for the target adjacent pixel group 303. This prediction signal is called a “predicted adjacent pixel group”. In FIG. 3, the pixel group 308 is acquired as the predicted adjacent pixel group 308 for the target adjacent pixel group 303. Then, the prediction signal generator 801 uses the pixel signal of the region 307 corresponding to the prediction adjacent pixel group 308 that has the same positional relationship as the positional relationship between the target adjacent pixel group 303 and the target block 302 as a prediction signal for the target block 302. Acquired as 307. Such a matching method is called template matching. As described above, the prediction signal for the target block can be identified by performing the matching using the inverted “L” target adjacent pixel group. Therefore, it is not necessary to encode motion information for specifying a prediction signal.

  Next, the parameter estimator 802 uses the target adjacent pixel group 303 and the predicted adjacent pixel group 308 acquired as described above to derive approximate parameters by the method described above. The derived approximate parameter is sent to the approximate signal generator 803, and the approximate signal generator 803 generates an approximate signal for the target block 302 by converting the prediction signal 307 using the approximate parameter by the method described above. The generated approximate signal is sent to the subtracter 105 in FIG. 1 via the line L103.

  In the above method, the reference image 306 different from the target image 301 is used as the reference region. However, the region 304 including the already reproduced pixels in the target image 301 or a part of the region 304 is used as the reference region, and the same as above. In-screen prediction can be realized by performing a matching process. In this case, the receiving side can identify the prediction signal for the target block by performing the same matching process. Therefore, it is not necessary to encode and transmit information related to the interpolation method for generating the prediction signal in the moving image predictive encoding device.

  FIG. 9 is a flowchart of a prediction process for generating an approximate signal according to the embodiment shown in FIG. In step 911 in FIG. 9, the prediction signal generator 801 obtains the target adjacent pixel group 303 adjacent to the target block 302 from the frame memory 104 via the line L104 based on the position information of the target block 302. The predicted neighboring pixel group 308 having the smallest error with the target neighboring pixel group 303 is acquired from the reference region by the template matching method. In the next step 912, the prediction signal generator 801 outputs the pixel signal of the region 307 corresponding to the prediction adjacent pixel group 308, which has the same positional relationship as that of the target adjacent pixel group 303 and the target block 302, for the target block 302. Obtained as a predicted signal 307.

  Next, in step 913, the parameter estimator 802 obtains an approximate parameter by the method described above using the target adjacent pixel group 303 and the predicted adjacent pixel group 308 acquired as described above. Specifically, the approximation parameters s and v are obtained using Equation (3) and Equation (4) that give the minimum value of Equation (2) described above.

The obtained approximate parameters s and v are sent to the approximate signal generator 803, and the approximate signal generator 803 converts the prediction signal as shown in Expression (6) using the approximate parameters s and v by the method described above. Thus, the approximate signal x ^* (i, j) for the target block 302 is generated (step 914). Finally, the approximate signal generator 803 outputs the generated approximate signal to the subtractor 105 in FIG. 1 (step 915).

  9 may be executed for all the target blocks in the encoding target image or may be executed for a part of the target blocks in the encoding target image. Further, the prediction signal converted in step 914 may be generated by the approximate signal generator 803 or may be generated by the prediction signal generator 801 in step 912 and sent to the approximate signal generator 803.

[About video predictive decoding apparatus]
Next, the video predictive decoding device according to the present embodiment will be described. FIG. 10 is a block diagram of the video predictive decoding device 1000 according to this embodiment. As shown in FIG. 10, the moving picture predictive decoding apparatus 1000 includes an input terminal 1001, a data analyzer 1002, an inverse quantizer 1003, an inverse transformer 1004, an adder 1005, a signal generator 1008, a frame memory 1007, and An output terminal 1006 is provided.

[Outline of operation in moving picture predictive decoding apparatus]
Hereinafter, the operation of each component of the moving picture predictive decoding apparatus 1000 will be described. The compressed data that has been compression-encoded by the moving picture predictive encoding apparatus 100 in FIG. 1 described above is input from the input terminal 1001. The compressed data includes information related to the generation of the encoded residual signal, the quantization parameter, and the approximate signal. Information related to the generation of the approximate signal includes at least one of the above-described motion information for inter-screen prediction and information related to an interpolation method for intra-screen prediction. Note that when an approximate signal is generated by template matching, information related to the generation of the approximate signal is not included.

  The data analyzer 1002 extracts the residual signal of the target block, information related to generation of the approximate signal, and the quantization parameter from the compressed data. Among these, the residual signal and quantization parameter of the target block are sent to the inverse quantizer 1003 via the line L1002, and the inverse quantizer 1003 inversely quantizes the residual signal of the target block based on the quantization parameter. . The residual signal after inverse quantization is sent to an inverse transformer 1004 via a line L1003, and the inverse transformer 1004 performs inverse discrete cosine transform on the residual signal after inverse quantization. As a result, the residual signal is restored, and the restored residual signal is sent to the adder 1005 via the line L1004.

  On the other hand, information related to the generation of the approximate signal is sent to the signal generator 1008 via the line L1002b, and the signal generator 1008 receives the information related to the generation of the approximate signal from the frame memory 1007 via the line L1007. A reference signal is acquired, and an approximate signal is generated by a method described later. This approximate signal is sent to the adder 1005 via the line L1008, and the adder 1005 adds the restored residual signal and the approximate signal to reproduce the target block signal. The reproduced target block signal is output to the outside from the output terminal 1006 via the line L1005 and stored in the frame memory 1007.

  Next, the configuration and operation of the signal generator 1008 provided in the video predictive decoding apparatus 1000 will be described with reference to FIG. As shown in FIG. 11, the signal generator 1008 includes a parameter estimator 1101 that estimates an approximate parameter, and an approximate signal generator 1102 that generates an approximate signal. Hereinafter, the operation of the signal generator 1008 will be described in the order of (1) inter-screen prediction and (2) intra-screen prediction.

[(1) Inter-screen prediction]
In the inter-screen prediction of the present embodiment, the parameter estimator 1101 provided in the signal generator 1008 estimates an approximate parameter for converting the prediction signal as follows. Specifically, the parameter estimator 1101 acquires a pixel group adjacent to the target block from the frame memory 1007 via the line L1007 based on the current position information of the target block to be decoded. The position information of the target block is managed by a control unit (not shown) of the entire system. In this embodiment, a pixel group adjacent to the target block is referred to as a “target adjacent pixel group”, and the target adjacent pixel group 303 corresponds to FIG.

  Next, the parameter estimator 1101 acquires a pixel group adjacent to the prediction signal specified by the motion information sent from the reference image via the line L1002b. This pixel group is called a “predicted adjacent pixel group”. The positional relationship between the predicted adjacent pixel group and the predicted signal is the same as the positional relationship between the target adjacent pixel group and the target block.

  Then, the parameter estimator 1101 derives approximate parameters using the acquired target adjacent pixel group x (i, j) and predicted adjacent pixel group y (i, j). In the present embodiment, approximate parameters s and v are obtained using Equation (3) and Equation (4) that give the minimum value of Equation (2) described above. These approximate parameters s and v are sent to the approximate signal generator 1102 via line L1101.

The approximate signal generator 1102 acquires a prediction signal corresponding to the target block from the frame memory 1007 based on the motion information sent via the line L1002b. Next, the approximate signal generator 1102 uses the approximate parameters s and v to convert the prediction signal as in Expression (6) to obtain each pixel x ^* (i, j) of the approximate signal. The approximate signal obtained in this way is sent to the adder 1005 in FIG. 10 via the line L1008.

[(2) In-screen prediction]
In the case of intra prediction, an approximate signal is generated by processing similar to inter prediction. Specifically, the parameter estimator 1101 obtains a pixel group (target adjacent pixel group) adjacent to the target block from the frame memory 1007 via the line L1007 based on the current position information of the target block to be decoded. To do. The position information of the target block is managed by a control unit (not shown) of the entire system. The target adjacent pixel group corresponds to, for example, the pixel group 602 in FIG.

  Next, the parameter estimator 1101 generates an interpolation signal for the target adjacent pixel group based on the information related to the interpolation method sent via the line L1002b. Specifically, the parameter estimator 1101 is an interpolation method specified by information regarding the interpolation method sent via the line L1002b, and uses pixels adjacent to the target adjacent pixel group (pixels 603 to 621 in FIG. 6). Then, an interpolation signal (predicted adjacent pixel group) corresponding to the target adjacent pixel group is generated. The generated predicted adjacent pixel group has the same shape as the target adjacent pixel group, and includes the same number of pixels.

  Further, the parameter estimator 1101 derives an approximate parameter using the predicted adjacent pixel group y (i, j) and the target adjacent pixel group x (i, j) generated in this way. In the present embodiment, approximate parameters s and v are obtained using Equation (3) and Equation (4) that give the minimum value of Equation (2) described above. These approximate parameters s and v are sent to the approximate signal generator 1102 via line L1101.

The approximate signal generator 1102 generates a prediction signal by using the already reproduced pixels around the target block by an interpolation method specified by information regarding the interpolation method input via the line L1002b. Specifically, the approximate signal generator 1102 obtains pixels (pixels A to M in FIG. 5) adjacent to the target block from the frame memory 1007 via the line L1007, and relates to the interpolation method sent via the line L1002b. A prediction signal is generated based on the information. Each pixel of the prediction signal is represented by y ^* (i, j) in Expression (6), and the prediction signal is converted as shown in Expression (6) using the approximation parameters s and v. Find x ^* (i, j). The obtained approximate signal is sent to the adder 1005 via the line L1008.

  FIG. 12 shows a flowchart of a prediction process performed by the signal generator 1008 to generate an approximate signal. Since this flowchart is common to the case of inter prediction and the case of intra prediction, both cases will be described below.

  In step 1211 of FIG. 12, the parameter estimator 1101 selects a pixel group (target adjacent pixel group) adjacent to the target block based on the position information of the current target block to be decoded via the line L1007 and the frame memory 1007. Get from. Next, in step 1212, the parameter estimator 1101 acquires or generates a signal of the predicted adjacent pixel group based on the identification information input from the outside. In the case of inter-screen prediction, the identification information is motion information, and the parameter estimator 1101 acquires a prediction adjacent pixel group based on the motion information. On the other hand, in the case of intra prediction, the identification information is information related to the interpolation method, and the parameter estimator 1101 generates a prediction adjacent pixel group based on the information related to the interpolation method. In the next step 1213, the parameter estimator 1101 obtains an approximate parameter from the target adjacent pixel group and the predicted adjacent pixel group by the method described above. The obtained approximate parameter is sent to the approximate signal generator 1102.

  In the next step 1214, the approximate signal generator 1102 acquires or generates a prediction signal based on the same identification information as in step 1212. In the case of inter-screen prediction, the identification information is motion information, and the approximate signal generator 1102 acquires a prediction signal based on the motion information. On the other hand, in the case of intra prediction, the identification information is information regarding the interpolation method, and the approximate signal generator 1102 generates a prediction signal based on the information regarding the interpolation method. Further, in step 1215, the approximate signal generator 1102 generates an approximate signal by converting the acquired or generated prediction signal using the approximate parameter obtained in step 1213. The generated approximate signal is output to the adder 1005 via the line L1008 (step 1216).

  2 in the video predictive coding apparatus corresponds to the signal generator 1008 in FIG. 11 in the video predictive decoding apparatus, and is a flowchart of signal generation processing (in the case of inter-screen prediction, FIG. In the case of intra prediction, FIG. 7) corresponds to the flowchart of FIG.

  By the way, another form of the signal generator 103 shown in FIG. 8 performs the same operation when it is provided in the moving picture predictive coding apparatus and when it is provided in the moving picture predictive decoding apparatus. Similarly, the flowchart of the generation process is the same for the encoding process and the decoding process. That is, the signal generator 103 provided in the moving picture predictive decoding apparatus performs the same operation as that provided in the above-described moving picture predictive encoding apparatus, and therefore, redundant description is omitted here.

  That is, the signal generator 103 of the video predictive decoding device on the receiving side can specify the prediction signal for the target block by performing the same matching process as the signal generator of the video predictive encoding device. Therefore, the video predictive decoding device does not need to receive information on the interpolation method for generating the prediction signal from the video predictive encoding device.

[Video Predictive Encoding Program / Video Predictive Decoding Program and Operating Environment]
By the way, the invention according to the video predictive coding apparatus can be understood as an invention according to a video predictive coding program for causing a computer to function as a video predictive coding apparatus. The invention relating to the moving picture predictive decoding apparatus can be regarded as an invention relating to a moving picture predictive decoding program for causing a computer to function as a moving picture predictive decoding apparatus. The moving picture predictive encoding program and the moving picture predictive decoding program are provided by being stored in a recording medium, for example. Examples of the recording medium include a recording medium such as a flexible disk, a CD-ROM, and a DVD, a semiconductor memory, and the like.

  FIG. 13 is a diagram illustrating a hardware configuration of the computer 30 for executing the program recorded on the recording medium 10, and FIG. 14 is a perspective view of the computer 30 for executing the program recorded on the recording medium 10. FIG. Here, the computer 30 includes a DVD player, a set top box, a mobile phone, and the like that have a CPU and perform processing and control by software.

  As shown in FIG. 13, the computer 30 includes a reading device 12 such as a flexible disk drive device, a CD-ROM drive device, and a DVD drive device, a working memory (RAM) 14 in which an operating system is resident, and a recording medium 10. A memory 16 for storing the program stored therein, a display 18, a mouse 20 and a keyboard 22 as input devices, a communication device 24 for transmitting and receiving data and the like, and a CPU 26 for controlling execution of the program. ing. When the recording medium 10 is inserted into the reading device 12, the computer 30 can access the moving image predictive encoding program and the moving image predictive decoding program stored in the recording medium 10 from the reading device 12. The encoding program and the video predictive decoding program allow the computer 30 to operate as the video predictive encoding device and the video predictive decoding device according to the present embodiment.

  As shown in FIG. 14, the moving picture predictive encoding program and the moving picture predictive decoding program may be provided as a computer data signal 40 superimposed on a carrier wave via a wired network / wireless network. In this case, the computer 30 may store the moving picture predictive encoding program and the moving picture predictive decoding program received by the communication device 24 in the memory 16 and execute the moving picture predictive encoding program and the moving picture predictive decoding program. it can.

  According to the present embodiment described above, the already-reproduced pixels that are in the same positional relationship as the positional relationship between the target adjacent pixel group and the target block with respect to the target adjacent pixel group including the already-reproduced pixels and the prediction block including the prediction signal. An approximate parameter for approximating the predicted adjacent pixel group to the target adjacent pixel group is determined using the predicted adjacent pixel group consisting of and the predicted signal is converted by using the approximate parameter to approximate the target block. Therefore, it is not necessary to store or transmit the approximate parameter after encoding it with the residual signal, and to suppress an increase in the amount of data due to the approximate parameter.

  Along with this, even when the image is divided into smaller blocks than before and luminance correction is performed on each small block, or when the target signal shows a non-linear change with respect to the prediction signal, the approximate parameter increases. It is possible to generate a high-precision approximate signal while avoiding an increase in the amount of auxiliary information data caused.

  As described above, a highly accurate approximate signal can be generated with a small amount of auxiliary information, the residual signal can be reduced, and the code amount of the entire moving image can be greatly reduced.

  The receiving side does not derive approximate parameters according to the above-described method, but includes related information (approximate parameters themselves or identification information for identifying predetermined approximate parameters) in the compressed data. After extracting the relevant information indicating the approximation parameter on the receiving side, the approximation signal is generated by applying the approximation parameter to the prediction signal generated according to the method specified by the interpolation method or the prediction signal determined by template matching. To do. In the case of the interpolation method shown in FIG. 5, the prediction signal may be generated after the approximation parameter is directly applied to the pixel group (A to M in FIG. 5) adjacent to the boundary of the target block.

  Also, after the moving picture predictive coding apparatus determines whether or not to apply the above approximation method to the prediction signal, the identification information “1” is applied when the approximation method is applied, and “0” when the approximation method is not applied. May be transmitted to the video predictive decoding apparatus. This identification information is transmitted and received in sequence units, frame units, or small area units. As a method for determining the identification information, for example, an error value A between the prediction signal and the signal of the target block, and an error value B between the approximate signal generated by applying the approximation parameter to the prediction signal and the signal of the target block, The identification information is set to “1” when the error value A is larger than the error value B, and the identification information is set to “0” when the error value B is larger than the error value A. Then, the moving picture predictive decoding apparatus generates an approximate signal based on the received identification information. That is, the moving picture predictive decoding apparatus generates an approximate signal by applying an approximate parameter after generating a prediction signal by the above-described method when the identification information is “1”, and approximate parameter when the identification information is “0”. The prediction signal is used as an approximate signal without applying the above.

It is a block diagram which shows the moving image predictive encoding apparatus of this embodiment. It is a block diagram which shows the signal generator used for a moving image predictive coding apparatus. It is a schematic diagram for demonstrating the production | generation of an approximate signal. It is a flowchart which shows the 1st process of the approximate signal generation which concerns on the inter-screen prediction performed with a moving image predictive coding apparatus. It is a schematic diagram which shows the specific method of producing | generating a prediction signal by the prediction in a screen. It is a schematic diagram for demonstrating the production | generation of the approximate signal in a screen. It is a flowchart which shows the 2nd process of the approximate signal generation which concerns on the prediction in a screen performed with a moving image predictive coding apparatus. It is a block diagram which shows another signal generator used for a moving image predictive coding apparatus and a moving image predictive decoding apparatus. FIG. 9 is a flowchart showing a third process for generating an approximate signal by template matching by the signal generator of FIG. 8. FIG. It is a block diagram which shows the moving image predictive decoding apparatus of this embodiment. It is a block diagram which shows the signal generator used for a moving image predictive decoding apparatus. It is a flowchart which shows the process of the approximate signal generation which concerns on the prediction between screens performed in a moving image prediction decoding apparatus, and the prediction in a screen. It is a figure which shows the hardware constitutions of the computer for performing the program recorded on the recording medium. It is a perspective view of a computer for executing a program stored in a recording medium.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 ... Recording medium, 12 ... Reading apparatus, 14 ... Working memory, 16 ... Memory, 18 ... Display, 20 ... Mouse, 22 ... Keyboard, 24 ... Communication apparatus, 30 ... Computer, 40 ... Computer data signal, 100 ... Movie Image predictive coding apparatus 101... Input terminal 102 102 block divider 103 signal generator 104 frame memory 105 subtractor 106 converter 107 quantizer 108 inverse quantizer , 109: inverse transformer, 110: adder, 111: entropy encoder, 112: output terminal, 201: prediction signal generator, 202 ... parameter estimator, 203 ... approximate signal generator, 801 ... prediction signal generator , 802 ... Parameter estimator, 803 ... Approximate signal generator, 1000 ... Moving picture predictive decoding apparatus, 1001 ... Input terminal, 1002 ... 1003 ... Inverse quantizer, 1004 ... Inverse transformer, 1005 ... Adder, 1006 ... Output terminal, 1007 ... Frame memory, 1008 ... Signal generator, 1101 ... Parameter estimator, 1102 ... Approximate signal generator .

Claims (10)

Area dividing means for dividing the input image into a plurality of areas;
A prediction signal generating unit that generates an approximate signal for a target pixel signal of a target region that is a processing target among a plurality of regions divided by the region dividing unit;
Residual signal generation means for generating a residual signal between the approximate signal generated by the prediction signal generation means and the target pixel signal;
Encoding means for encoding the residual signal generated by the residual signal generating means;
With
The prediction signal generation means determines a prediction signal for the target region by a predetermined method, and a target adjacent pixel group including already reproduced pixels adjacent to the target region, and a reference region including the prediction signal, Approximate parameter for approximating the predicted adjacent pixel group to the target adjacent pixel group using a predicted adjacent pixel group consisting of already reproduced pixels that has the same positional relationship as the positional relationship between the target region and the target adjacent pixel group And generating an approximate signal for the target pixel signal of the target region by converting the prediction signal for the target region based on the approximate parameter.
A video predictive coding apparatus characterized by the above. Area dividing means for dividing the input image into a plurality of areas;
A prediction signal generating unit that generates an approximate signal for a target pixel signal of a target region that is a processing target among a plurality of regions divided by the region dividing unit;
Residual signal generation means for generating a residual signal between the approximate signal generated by the prediction signal generation means and the target pixel signal;
Encoding means for encoding the residual signal generated by the residual signal generating means;
With
The prediction signal generation means uses a region composed of pixels reproduced before the target region as a reference region, and applies a reference region of the reference region to a target adjacent pixel group composed of already reproduced pixels adjacent to the target region. A pixel group having the highest correlation among them is derived as a predicted adjacent pixel group, and a pixel signal of a reference region corresponding to the predicted adjacent pixel group having the same positional relationship as the positional relationship between the target adjacent pixel group and the target region is obtained. , Obtaining as a prediction signal for the target region, determining an approximate parameter for approximating the predicted adjacent pixel group to the target adjacent pixel group using the predicted adjacent pixel group and the target adjacent pixel group, By generating the approximate signal for the target pixel signal of the target region by converting the prediction signal for the target region based on the approximate parameter,
A video predictive coding apparatus characterized by the above. Data analysis means for extracting encoded data of the residual signal for the target region to be processed from the compressed data;
Residual signal restoration means for restoring a reproduction residual signal from encoded data of the residual signal extracted by the data analysis means;
Predicted signal generating means for generating an approximate signal for the target pixel signal of the target region;
Image restoration means for restoring the target pixel signal of the target area by adding the approximate signal generated by the prediction signal generation means and the reproduction residual signal restored by the residual signal restoration means;
With
The prediction signal generation unit obtains a prediction signal for the target region by a predetermined method, and a target adjacent pixel group composed of already reproduced pixels adjacent to the target region, and a reference region including the prediction signal, Approximate parameter for approximating the predicted adjacent pixel group to the target adjacent pixel group using a predicted adjacent pixel group consisting of already reproduced pixels that has the same positional relationship as the positional relationship between the target region and the target adjacent pixel group And generating an approximate signal for the target pixel signal of the target region by converting the prediction signal for the target region based on the approximate parameter.
A video predictive decoding apparatus characterized by the above. Data analysis means for extracting encoded data of the residual signal for the target region to be processed from the compressed data;
Residual signal restoration means for restoring a reproduction residual signal from encoded data of the residual signal extracted by the data analysis means;
Predicted signal generating means for generating an approximate signal for the target pixel signal of the target region;
Image restoration means for restoring the target pixel signal of the target area by adding the approximate signal generated by the prediction signal generation means and the reproduction residual signal restored by the residual signal restoration means;
With
The prediction signal generation means uses a region composed of pixels reproduced before the target region as a reference region, and applies a reference region of the reference region to a target adjacent pixel group composed of already reproduced pixels adjacent to the target region. An approximate parameter for deriving the pixel group having the highest correlation from among the predicted adjacent pixel groups and approximating the predicted adjacent pixel group to the target adjacent pixel group using the predicted adjacent pixel group and the target adjacent pixel group And converting the pixel signal of the reference region corresponding to the predicted adjacent pixel group, which is in the same positional relationship as the positional relationship between the target adjacent pixel group and the target region, based on the approximation parameter, Generate an approximate signal for the region,
A video predictive decoding apparatus characterized by the above. A video predictive encoding method executed by a video predictive encoding device,
A region dividing step for dividing the input image into a plurality of regions;
A prediction signal generating step for generating an approximate signal for a target pixel signal of a target region that is a processing target among a plurality of regions divided by the region dividing step;
A residual signal generation step for generating a residual signal between the approximate signal generated by the prediction signal generation step and the target pixel signal;
An encoding step for encoding the residual signal generated by the residual signal generation step;
With
In the prediction signal generation step, a prediction signal for the target region is determined by a predetermined method, and a target adjacent pixel group including already reproduced pixels adjacent to the target region and a reference region including the prediction signal, Approximate parameter for approximating the predicted adjacent pixel group to the target adjacent pixel group using a predicted adjacent pixel group consisting of already reproduced pixels that has the same positional relationship as the positional relationship between the target region and the target adjacent pixel group And generating an approximate signal for the target pixel signal of the target region by converting the prediction signal for the target region based on the approximate parameter.
A video predictive encoding method characterized by the above. A video predictive encoding method executed by a video predictive encoding device,
A region dividing step for dividing the input image into a plurality of regions;
A prediction signal generating step for generating an approximate signal for a target pixel signal of a target region that is a processing target among a plurality of regions divided by the region dividing step;
A residual signal generation step for generating a residual signal between the approximate signal generated by the prediction signal generation step and the target pixel signal;
An encoding step for encoding the residual signal generated by the residual signal generation step;
With
In the predicted signal generation step, an area composed of pixels reproduced before the target area is set as a reference area, and a target adjacent pixel group consisting of already reproduced pixels adjacent to the target area is compared with the reference area. A pixel group having the highest correlation among them is derived as a predicted adjacent pixel group, and a pixel signal of a reference region corresponding to the predicted adjacent pixel group having the same positional relationship as the positional relationship between the target adjacent pixel group and the target region is obtained. , Obtaining as a prediction signal for the target region, determining an approximate parameter for approximating the predicted adjacent pixel group to the target adjacent pixel group using the predicted adjacent pixel group and the target adjacent pixel group, By generating the approximate signal for the target pixel signal of the target region by converting the prediction signal for the target region based on the approximate parameter,
A video predictive encoding method characterized by the above. A video predictive decoding method executed by a video predictive decoding device,
A data analysis step for extracting encoded data of the residual signal for the target region to be processed from the compressed data;
A residual signal restoration step for restoring a reproduction residual signal from the encoded data of the residual signal extracted by the data analysis step;
A prediction signal generation step of generating an approximate signal for the target pixel signal of the target region;
An image restoration step of restoring the target pixel signal of the target region by adding the approximate signal generated by the prediction signal generation step and the reproduction residual signal restored by the residual signal restoration step;
With
In the prediction signal generation step, a prediction signal for the target region is obtained by a predetermined method, and the target adjacent pixel group including already reproduced pixels adjacent to the target region, and the reference region including the prediction signal, Approximate parameter for approximating the predicted adjacent pixel group to the target adjacent pixel group using a predicted adjacent pixel group consisting of already reproduced pixels that has the same positional relationship as the positional relationship between the target region and the target adjacent pixel group And generating an approximate signal for the target pixel signal of the target region by converting the prediction signal for the target region based on the approximate parameter.
A video predictive decoding method characterized by the above. A video predictive decoding method executed by a video predictive decoding device,
A data analysis step for extracting encoded data of the residual signal for the target region to be processed from the compressed data;
A residual signal restoration step for restoring a reproduction residual signal from the encoded data of the residual signal extracted by the data analysis step;
A prediction signal generation step of generating an approximate signal for the target pixel signal of the target region;
An image restoration step of restoring the target pixel signal of the target region by adding the approximate signal generated by the prediction signal generation step and the reproduction residual signal restored by the residual signal restoration step;
With
In the predicted signal generation step, an area composed of pixels reproduced before the target area is set as a reference area, and a target adjacent pixel group consisting of already reproduced pixels adjacent to the target area is compared with the reference area. An approximate parameter for deriving the pixel group having the highest correlation from among the predicted adjacent pixel groups and approximating the predicted adjacent pixel group to the target adjacent pixel group using the predicted adjacent pixel group and the target adjacent pixel group And converting the pixel signal of the reference region corresponding to the predicted adjacent pixel group, which is in the same positional relationship as the positional relationship between the target adjacent pixel group and the target region, based on the approximation parameter, Generate an approximate signal for the region,
A video predictive decoding method characterized by the above.   A moving picture predictive coding program for causing a computer to function as the moving picture predictive coding apparatus according to claim 1. A moving picture predictive decoding program for causing a computer to function as the moving picture predictive decoding apparatus according to claim 3 or 4.

Images