JP2017228827A - Intra-prediction device, image coding device, image decoding device, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the encoding efficiency of an intra-prediction mode.SOLUTION: An error index acquisition unit acquires an error index between a predicted image obtained by using an intra-prediction mode determining about a reference block around an object block and an input image. An order determination unit determines an ascending order of error indexes as the order of intra-prediction mode candidates, when using the intra-prediction mode determining the reference block as the intra-prediction mode of the object block.SELECTED DRAWING: Figure 1

Description

  The present invention relates to an intra predictor, an image encoding device, an image decoding device, and a program.

  As one of the typical video compression methods, MPEG-H HEVC / ITU-TH. H265 (Moving Picture Expert Group-H ISO / IEC 23008 High Efficiency Video Coding / International Telecommunication Union-Telecommunication Sect. In the HEVC scheme, a group of intra prediction mode candidates called MPM (Most Probable Mode) is used in intra prediction.

  The HEVC MPM is configured for each intra prediction target block, and includes three intra prediction mode candidates (hereinafter referred to as intra prediction mode candidates). In principle, the 0th of the three intra prediction mode candidates in the MPM is the intra prediction mode of the left reference block (hereinafter referred to as the left block) of the target block, and the first is the reference block above the target block (hereinafter, the right block). Intra prediction mode (referred to as a block), and the second is defined as a planar mode. However, in cases where the left block or the upper block cannot be used or the intra prediction mode of the left block matches the intra prediction mode of the right block, the intra prediction mode candidates in the MPM and their ranks are determined. ing.

  In the HEVC scheme, the intra prediction mode encoded stream includes (a) a flag indicating whether or not the intra prediction mode in the MPM is used, and (b) whether or not the 0th intra prediction mode candidate in the MPM is used. And (c) a flag indicating whether or not the first intra prediction mode candidate in the MPM is used. In the encoded stream, flags (a) to (c) are arranged in that order. The flag (b) and the flag (c) exist only when the value of the flag (a) is true, and the flag (c) exists only when the value of the flag (b) is false. Therefore, when the intra prediction mode candidate with the rank 0 in the MPM is used among the intra prediction mode candidates in the MPM, the flag (a) and the flag (b) are included in the encoded stream. That is, the information amount in the intra prediction mode is 2 bits. On the other hand, when the first or second intra prediction mode candidate is used, flags (a), (b), and (c) are all included in the encoded stream. included. That is, 3 bits of information is required for encoding in the intra prediction mode. Therefore, when the intra prediction mode is encoded using the MPM, the rank in the MPM of the intra prediction mode candidate (hereinafter referred to as MPM rank) affects the amount of information generated in the encoded stream generated by the encoding. .

JP2013-081100A

  In the conventional encoding method, for example, the HEVC method or the encoding process described in Patent Document 1, the rank of the MPM is determined based on the spatial positional relationship of the intra prediction mode candidates. Therefore, the conventional coding method has a problem that the coding efficiency of the intra prediction mode is not necessarily increased.

  The present invention has been made in view of the above points, and an object thereof is to provide an intra predictor, an image encoding device, an image decoding device, and a program capable of improving the encoding efficiency of the intra prediction mode. .

The present invention has been made to solve the above problems. [1] According to one aspect of the present invention, a prediction image obtained by using an intra prediction mode defined for reference blocks around a target block and an input are input. When using an error index acquisition unit for acquiring an error index with respect to an image and an intra prediction mode defined for the reference block as an intra prediction mode candidate of the target block, the rank of the intra prediction mode candidates is increased in ascending order of the error index. An intra predictor comprising:
According to the configuration of [1], there is a high possibility that the representative intra prediction mode that gives the predicted image with the smallest amount of error from the input image matches the intra prediction mode candidate with the higher rank. Therefore, it is possible to reduce the number of intra prediction mode candidates related to encoding in the intra prediction mode, that is, the number of encoded bits.

[2] One aspect of the present invention is the intra predictor described above, wherein the rank determination unit includes the reference block that gives the error index larger than a predetermined error index threshold as the intra prediction mode candidate. The intra-prediction mode rank is lower than a predetermined intra-prediction mode candidate rank.
According to the configuration of [2], the rank of the intra prediction mode candidate in the intra prediction mode related to the reference pixel that gives an error index larger than the predetermined error index threshold is lower than that of the predetermined intra prediction mode candidate. Therefore, in the intra prediction mode encoding, the intra prediction mode candidate having a smaller error index is preferentially used. Therefore, it is possible to reduce the number of intra prediction mode candidates used in encoding the intra prediction mode related to the target block, that is, the number of encoded bits, rather than only the intra prediction mode related to the reference block.

[3] One aspect of the present invention is the intra predictor described above, wherein the error index acquisition unit sets a threshold value of the error index based on a range of an error index related to an intra prediction mode determined for the reference block. It is characterized by defining.
According to the configuration of [3], in encoding of the intra prediction mode, a predetermined intra prediction mode candidate having a smaller error index than the intra prediction mode related to the reference pixel is prioritized based on a predetermined error index threshold. It is determined whether or not to use. Therefore, the number of encoded bits in the intra prediction mode is smaller than the order of the intra prediction mode of the reference block that always gives an error index larger than a certain error index threshold than the rank of the predetermined intra prediction mode candidate. can do.

[4] One aspect of the present invention is the intra predictor described above, wherein the error index acquisition unit obtains a quantization transform coefficient of a prediction residual based on the prediction image as the error index by variable length encoding. The number of bits of the bit string to be obtained is obtained.
According to the configuration of [4], the ranking of the intra prediction mode candidates is determined in ascending order of the number of bits of the bit string of the prediction residual based on the intra prediction image generated using each intra prediction mode candidate. Intra prediction mode candidates with a small number of bits as the size of the prediction residual are more likely to match the intra prediction mode related to the target block, so the number of coding bits related to the coding of the intra prediction mode is reduced. be able to.

[5] One aspect of the present invention is the above-described intra predictor, wherein the error index acquisition unit sets a value other than zero among the quantized transform coefficients of a prediction residual based on the predicted image as the error index. The number of non-zero transform coefficients is obtained.
According to this configuration, the intra prediction mode candidates are ranked with a small number of non-zero transform coefficients of the prediction residual based on the intra prediction image generated using each intra prediction mode candidate. There is a high possibility of matching with the intra prediction mode related to the block. Therefore, it is possible to reduce the number of encoded bits related to the intra prediction mode encoding.

[6] One aspect of the present invention is the intra predictor described above, wherein the error index acquisition unit acquires a square sum of the quantized transform coefficients of a prediction residual based on the predicted image as the error index.
According to the configuration of [6], an intra prediction mode in which the ranks of intra prediction mode candidates are small in the sum of squares of quantized transform coefficients of prediction residuals based on intra prediction images generated using the respective intra prediction mode candidates. There is a high possibility that the candidate matches the intra prediction mode related to the target block. Therefore, it is possible to reduce the number of encoded bits related to encoding in the intra prediction mode.

[7] One aspect of the present invention is the intra predictor described above, wherein the error index acquisition unit acquires a sum of absolute values of quantized transform coefficients of a prediction residual based on the predicted image as the error index. .
According to the configuration of [7], the intra prediction mode candidate rank is an intra prediction with a small absolute value sum of quantized transform coefficients of a prediction residual based on an intra prediction image generated using each intra prediction mode candidate. There is a high possibility that the mode candidate matches the intra prediction mode related to the target block. Therefore, it is possible to reduce the number of encoded bits related to encoding in the intra prediction mode.

[8] One aspect of the present invention includes the intra predictor according to any one of [1] to [7], and performs intra prediction using each predetermined intra prediction mode with reference to a decoded image of the reference block. Candidate determination that selects one of the intra prediction modes based on an error amount between the predicted image obtained in this way and the input image of the target block, and determines the rank of the same intra prediction mode candidate as the selected intra prediction mode And an image encoding device.
According to the configuration of [8], the ranking of the same intra prediction mode candidate as the intra prediction mode selected based on the error amount between the predicted image and the input image becomes higher. Since the number of intra prediction mode candidates for encoding can be reduced, the number of encoded bits in the intra prediction mode can be reduced.

[9] An aspect of the present invention includes an intra predictor according to any one of [1] to [7], and a predicted image obtained using the same intra prediction mode as the intra prediction mode candidate of the input rank The image decoding apparatus is characterized in that a decoded image of the target block is generated using.
According to the configuration of [9], a decoded image is generated using a prediction image based on the same intra prediction mode as the intra prediction mode candidate of the input rank among the intra prediction mode candidates ordered in ascending order of the error index. Is done. Therefore, it is possible to decode intra prediction modes encoded in a higher order with a smaller number of encoded bits.

[10] One aspect of the present invention is an error index acquisition unit that acquires an error index between a predicted image and an input image obtained using an intra prediction mode determined for reference blocks around a target block in a computer, When the intra prediction mode defined for the reference block is used as the intra prediction mode candidate of the target block, the intra block predictor is made to function as an intra predictor including a rank determining unit that ranks the intra prediction mode candidates in ascending order of the error index. It is a program for.
According to the configuration of [10], there is a high possibility that the representative intra prediction mode that gives the predicted image with the smallest amount of error from the input image matches the intra prediction mode candidate with the higher rank. Therefore, it is possible to reduce the number of intra prediction mode candidates related to encoding in the intra prediction mode, that is, the number of encoded bits.

  According to the present invention, it is possible to improve the encoding efficiency of the intra prediction mode.

It is a schematic block diagram which shows the structure of the image coding apparatus which concerns on 1st Embodiment. It is a schematic block diagram which shows the structure of the image decoding apparatus which concerns on 1st Embodiment. It is a schematic block diagram which shows the structure of the intra estimation part which concerns on 1st Embodiment. It is a figure which shows the example of an object block and its reference block. It is a figure which shows the example of intra prediction mode. It is a schematic block diagram which shows the structure of the intra estimation part which concerns on 1st Embodiment. It is a flowchart which shows the MPM determination process which concerns on 1st Embodiment. It is a flowchart which shows the MPM determination process which concerns on 2nd Embodiment.

(First embodiment)
Hereinafter, a first embodiment of the present invention will be described with reference to the drawings.
A configuration of the image encoding device 1 according to the present embodiment will be described. FIG. 1 is a schematic block diagram illustrating a configuration of an image encoding device 1 according to the present embodiment. In the example illustrated in FIG. 1, the image encoding device 1 applies an intra prediction unit 110 (described later) as an example of an intra predictor according to the present embodiment to an image encoding device that performs image encoding processing according to the HEVC scheme. Do it.

  The image encoding device 1 includes a subtraction unit 101, an orthogonal transformation unit 102, a quantization unit 103, a variable length coding unit 104, an inverse quantization unit 105, an inverse orthogonal transformation unit 106, an addition unit 107, a loop filter 108, and a reference memory. 109, an intra prediction unit 110, an inter prediction unit 111, and a selection unit 112.

  An input image for each picture is input to the subtraction unit 101 from the outside of the image encoding device 1. A picture means one image and is also called a frame. A prediction image in each target block focused on as a processing target at that time is input from the selection unit 112 to the subtraction unit 101. Hereinafter, the predicted image in the target block is referred to as a predicted image block. The target block is an area having a predetermined size obtained by dividing a target picture that is a picture of interest as a processing target at that time. The size of the block is called the block size. The block size is, for example, any of 32 × 32 pixels, 16 × 16 pixels, 8 × 8 pixels, and 4 × 4 pixels. The subtraction unit 101 generates a residual image block by subtracting a predicted image block from an input image block that is an input image in the target block for each block, and outputs the generated residual image block to the orthogonal transform unit 102.

  The orthogonal transform unit 102 performs orthogonal transform on the residual image block input from the subtraction unit 101 to calculate an orthogonal transform coefficient, and outputs the calculated orthogonal transform coefficient to the quantization unit 103. As the orthogonal transform, for example, a discrete cosine transform (DCT) or a discrete sine transform (DST) is used.

  The quantization unit 103 quantizes the orthogonal transform coefficient input from the orthogonal transform unit 102 using a predetermined quantization parameter, and calculates a quantized orthogonal transform coefficient. The quantization unit 103 outputs a code indicating the calculated quantized orthogonal transform coefficient to the variable length coding unit 104 and the inverse quantization unit 105.

  The variable length coding unit 104 aggregates the quantization orthogonal transform coefficient code input from the quantization unit 103, representative intra prediction mode information input from the intra prediction unit 110, and motion information input from the inter prediction unit 111. To generate a data string. The variable length coding unit 104 performs variable length coding processing on the generated data string. The variable length coding process is also called entropy coding. The amount of information of various types of input data is compressed by the variable length encoding process. For example, a context-adaptive binary arithmetic coding (CABAC) is used as the variable-length coding process. The variable length encoding unit 104 outputs encoded data including a bit string obtained by performing variable length encoding processing to the outside of the image encoding device 1. The prediction mode information from the intra prediction unit 110 will be described later.

  The inverse quantization unit 105 performs inverse quantization on the code of the quantized orthogonal transform coefficient input from the quantization unit 103 using a predetermined quantization parameter to calculate an inverse quantized orthogonal transform coefficient. The inverse quantization unit 105 outputs the calculated inverse quantization orthogonal transform coefficient to the inverse orthogonal transform unit 106.

  The inverse orthogonal transform unit 106 performs inverse orthogonal transform on the inverse quantized orthogonal transform coefficient input from the inverse quantization unit 105, and generates a decoded residual image block that is a decoded residual image in the target block. The inverse orthogonal transform is an inverse operation with the orthogonal transform performed by the orthogonal transform unit 102. The inverse orthogonal transform unit 106 outputs the decoded residual image block to the addition unit 107.

  The addition unit 107 adds the decoded residual image block to the prediction image block of the target block input from the selection unit 112, and generates a decoded image block that is a decoded image in the target block. The addition unit 107 outputs the generated decoded image block to the loop filter 108 and the intra prediction unit 110. The decoded image block corresponds to a decoded image block obtained by performing the encoding process.

The loop filter 108 includes an element filter that performs a predetermined filter process on the decoded image block input from the adding unit 107, and sequentially stores the decoded image block obtained by the filter process in the reference memory 109. Therefore, the reference memory 109 stores a decoded image of a picture at each time point within a predetermined time from the target picture including the target picture.
The loop filter 108 includes a deblocking filter (DF: Deblocking Filter) and a pixel adaptive offset (SAO) as element filters.

  The intra prediction unit 110 receives a decoded image block from the addition unit 107 for each block. The intra prediction unit 110 performs intra prediction for each of a plurality of predetermined intra prediction modes with reference to a decoded image in a reference block around the target picture. As the reference block, a decoded image block that has been decoded within a predetermined range from the target block is used. The intra prediction unit 110 performs intra prediction for each intra prediction mode to generate a predicted image in the target block. Hereinafter, the prediction image in the target block is referred to as an intra prediction block. The intra prediction unit 110 outputs the generated intra prediction block to the selection unit 112.

  The intra prediction unit 110 calculates an error amount between the intra prediction image block generated for each intra prediction mode and the input image block. The intra prediction unit 110 determines an intra prediction mode having the smallest error amount acquired from a plurality of intra prediction modes as a representative intra prediction mode. In addition, the intra prediction unit 110 acquires a representative intra prediction mode determined for the reference block and its error index. Here, the error index is an index value indicating the magnitude of the residual that is the difference between the input image block and the predicted image block in the reference block. The error index is distinguished from an error amount indicating the magnitude of the residual that is the difference between the input image block and the predicted image block in the target block. When using the representative intra prediction mode defined for the reference block as the intra prediction mode candidate of the target block, the intra prediction unit 110 determines the rank of the intra prediction mode candidates in ascending order of the error index. Then, the intra prediction unit 110 determines whether or not there is an intra prediction mode candidate that matches the representative intra prediction mode related to the target block. The intra prediction unit 110 generates representative intra prediction mode information including information on the ranks of intra prediction mode candidates determined to match, and outputs the generated representative intra prediction mode information to the variable-length encoding unit 104. The configuration of the intra prediction unit 110 will be described later.

  The inter prediction unit 111 determines a motion vector by performing block matching on a target block of an input image input from the outside with reference to a decoded image of a picture within a predetermined period stored in the reference memory 109. Pictures within a predetermined period include past and future pictures than the target picture. In block matching, the inter prediction unit 111 calculates, for each displacement amount, an error amount with respect to an input image block of an inter prediction image block obtained based on a decoded image within a predetermined range from the target block. The inter prediction unit 111 determines a displacement amount having the smallest error amount as a motion vector. The inter prediction unit 111 performs motion compensation on the decoded image in the reference block using the determined motion vector, and generates a predicted image in the target block. Hereinafter, the prediction image in the target block is referred to as an inter prediction image block. The inter prediction unit 111 outputs the generated inter prediction image block to the selection unit 112. The inter prediction unit 111 outputs motion information including information on the determined motion vector to the variable length coding unit 104. Note that the motion vector defined for the target block may be equal to the motion vector defined for any of the encoded reference blocks. In that case, the inter prediction unit 111 identifies a reference block that gives a motion vector equal to the motion vector determined for the target block. Then, the inter prediction unit 111 includes the position information of the identified reference block in the motion information instead of the motion vector information.

  The selection unit 112 switches between the intra prediction image block input from the intra prediction unit 110 and the inter prediction image block input from the inter prediction unit 111, and the prediction image block selected by switching is switched to the subtraction unit 101 and the addition unit 107. Output.

  Next, the configuration of the image decoding device 2 according to the present embodiment will be described. FIG. 2 is a schematic block diagram illustrating a configuration of the image decoding device 2 according to the present embodiment. In the example illustrated in FIG. 1, the image decoding device 2 is configured by applying an intra prediction unit 207 (described later) as another example of the intra predictor according to the present embodiment to the image decoding device according to the HEVC scheme.

  The image decoding device 2 includes a variable length decoding unit 201, an inverse quantization unit 202, an inverse orthogonal transform unit 203, an addition unit 204, a loop filter 205, a reference memory 206, an intra prediction unit 207, and an inter prediction unit 208. The The processes performed by the inverse quantization unit 202, the inverse orthogonal transform unit 203, and the addition unit 204 are the same as those performed by the inverse quantization unit 105, the inverse orthogonal transform unit 106, and the addition unit 107 of the image encoding device 1, respectively.

  The variable length decoding unit 201 receives encoded data generated by the image encoding device 1 from the outside of the image decoding device 2. The variable length decoding unit 201 performs variable length decoding on a bit string included in input encoded data, and obtains a code of a quantized orthogonal transform coefficient, representative intra prediction mode information, and motion information for each block. The variable length decoding method is a variable length decoding method corresponding to the variable length coding method used for generating encoded data. The variable length decoding unit 201 outputs the code of the quantized orthogonal transform coefficient, the representative intra prediction mode information, and the motion information related to each target block to the inverse quantization unit 202, the intra prediction unit 207, and the inter prediction unit 208, respectively.

  The inverse quantization unit 202 inversely quantizes the quantization orthogonal transform coefficient code of the target block input from the variable length decoding unit 201 using a predetermined quantization parameter, and calculates the inverse quantization orthogonal transform coefficient. To do. The inverse quantization unit 202 outputs the calculated inverse quantization orthogonal transform coefficient to the inverse orthogonal transform unit 203.

  The inverse orthogonal transform unit 203 performs inverse orthogonal transform on the inverse quantized orthogonal transform coefficient of the target block input from the inverse quantization unit 202 to generate a decoded residual image block. The inverse orthogonal transform unit 203 outputs the decoded residual image block to the addition unit 204.

  The intra prediction image block related to the target block from the intra prediction unit 207 or the inter predicted image block related to the target block from the inter prediction unit 208 is input to the addition unit 204 as a predicted image block. The adding unit 204 adds the predicted image block and the decoded residual image block input from the inverse orthogonal transform unit 203 to generate a decoded image block. The adding unit 204 outputs the generated decoded image block to the outside of the loop filter 205 and the image decoding device 2.

The loop filter 205 includes DF and SAO as element filters for performing predetermined filter processing on the decoded image block related to the target block input from the adding unit 204, and sequentially stores the decoded image blocks obtained by the filter processing in the reference memory 206. Remember. Therefore, the reference memory 206 stores the decoded image of the picture at each time point including the target picture.
The intra prediction unit 207 receives the representative intra prediction mode information of the target block from the variable length decoding unit 201 and the decoded image block from the addition unit 204.

The intra prediction unit 207 acquires a representative intra prediction mode and its error index determined for a decoded reference block within a predetermined range from the target picture. The intra prediction unit 207
When the representative intra prediction mode defined for the reference block is used as the intra prediction mode candidate of the target block, the ranking of the intra prediction mode candidates is determined in ascending order of the error index.
Of the intra prediction mode candidates, the intra prediction unit 207 identifies the intra prediction mode candidates of the rank indicated by the representative intra prediction mode information input from the variable length decoding unit 201 as the representative intra prediction mode. The intra prediction unit 207 refers to the decoded image block of the reference block input from the addition unit 204, and outputs an intra prediction image block generated by performing intra prediction using the identified intra prediction mode to the addition unit 204. . The configuration of the intra prediction unit 207 will be described later.

  The inter prediction unit 208 receives motion information related to the target block from the variable length decoding unit 201. When the input motion information indicates a motion vector, the inter prediction unit 208 performs motion compensation on the decoded image within the predetermined range from the target block using the motion vector. When the input motion information indicates the position information of the reference block, the inter prediction unit 208 performs motion compensation using the motion vector of the reference block specified by the position information. The inter prediction unit 208 outputs an inter prediction image block obtained by performing motion compensation to the addition unit 204.

(Intra prediction unit)
Next, the configuration of the intra prediction unit 110 of the image encoding device 1 according to the present embodiment will be described. FIG. 3 is a schematic block diagram illustrating a configuration of the intra prediction unit 110 according to the present embodiment. The intra prediction unit 110 includes an intra predictor 1100 and a candidate determination unit 1107. The intra predictor 1100 includes an intra predicted image generation unit 1101, a subtraction unit 1102, an MPM determination unit 1103, and an intra prediction mode information generation unit 1106.

  The intra prediction image generation unit 1101 performs intra prediction for each of a plurality of predetermined intra prediction modes with reference to a decoded image in a reference block within a predetermined range from the target block. In the HEVC scheme, a block that is a processing unit of intra prediction corresponds to a processing unit (TU: Transform Unit) in which orthogonal transformation and inverse orthogonal transformation are performed. The number of intra prediction modes is 35 modes. A specific example of the intra prediction mode will be described later. The intra predicted image generation unit 1101 outputs the intra predicted image block obtained by the intra prediction to the subtraction unit 1102 and the candidate determination unit 1107.

  The subtraction unit 1102 subtracts the intra prediction image block input from the intra prediction image generation unit 1101 from the input image block, and generates a residual image block for each intra prediction mode. The subtraction unit 1102 outputs the generated residual image block to the candidate determination unit 1107.

  The MPM determination unit 1103 determines the MPM and MPM rank based on the error index of each decoded reference block and the representative intra prediction mode. The MPM is a group of intra prediction mode candidates that are a part of a plurality of intra prediction modes. The MPM rank is a rank of intra prediction mode candidates belonging to the MPM. In the HEVC method, as shown in FIG. 4, when determining the MPM, the block A on the left side of the target block T and the block B on the upper side of the target block are used. The MPM determination unit 1103 includes an error index acquisition unit 1104 and an MPM rank determination unit 1105.

  The error index acquisition unit 1104 acquires the error indexes Da and Db of the reference blocks A and B from a storage unit (not shown) included in the intra predictor 1100. The reference blocks A and B are blocks adjacent to the left and top of the target block T. The error index is an index value estimated to show a tendency similar to the error amount of the residual image related to the target block. In this embodiment, an index value that quantitatively represents the magnitude of the residual in each decoded reference block adjacent to the target block is used as an error index. Since the error index is derived from the decoded residual image block, it can be acquired by both the intra prediction unit 110 of the image encoding device 1 and the intra prediction unit 207 of the image decoding device 2. The residual image block is obtained by performing intra prediction for each of the reference blocks A and B using the representative intra prediction modes Pa and Pb. The representative intra prediction mode that gives the smallest error index in the reference block is presumed to be a highly reliable intra prediction mode that can faithfully reproduce the input image even in the target block adjacent to the reference block. Because.

  As the error index, for example, any of the following types of index values may be used. (1) Number of bits of a bit string obtained by variable length coding a code indicating a quantized orthogonal transform coefficient, (2) Number of non-zero transform coefficients having a non-zero value among quantized orthogonal transform coefficients, (3) DC component of quantized orthogonal transform coefficient, (4) sum of squares of quantized orthogonal transform coefficient, and (5) absolute value sum of quantized orthogonal transform coefficient. In the image encoding device 1, (1) and (2) can be acquired from the variable length encoding unit 104 and the inverse quantization unit 105, respectively. Any of (3) to (5) can be acquired from the inverse orthogonal transform unit 106.

As for (1), when the orthogonal transform coefficient becomes smaller, the quantized orthogonal transform coefficient becomes 0 or an integer value approximated to 0, and the bits that take a value of 0 by binarization are mainly used. Therefore, it is estimated that the smaller the value of (1), the smaller the prediction residual due to the same intra prediction mode as the representative intra prediction mode in the target block.
Regarding (2), when the orthogonal transform coefficient to be encoded becomes smaller, the quantized orthogonal transform coefficient becomes 0 or an integer value approximated to 0. Therefore, it is estimated that the smaller the value of (2) counted, the smaller the prediction residual in the target block.
(3) corresponds to the average value within the target block of pixel values for each pixel. It is estimated that the smaller the value of (3), the smaller the prediction residual in the same intra prediction mode as that of the representative intra prediction mode in the target block.
(4) corresponds to the sum in the target block of the square value of the quantized orthogonal transform coefficient for each spatial frequency. It is presumed that the smaller the value of (4), the smaller the prediction residual due to the same intra prediction mode as the representative intra prediction mode in the target block.
(5) corresponds to the sum in the target block of the absolute value of the quantized orthogonal transform coefficient for each spatial frequency. It shows that the prediction residual by the intra prediction mode same as the said representative intra prediction mode in the object block is so small that the value of (5) is small.

  The error index acquisition unit 1104 uses, as an error index, the error amount acquired by the candidate determination unit 1107 in the process of determining the representative intra prediction mode for the past target block, that is, the reference block, or the index value acquired in the process of calculating the error amount. It may be adopted. Therefore, it is not necessary to perform an additional process for calculating an error index related to the representative intra prediction mode of the reference block. Note that for the representative intra prediction mode of an invalid block, the error index acquisition unit 1104 determines a value Z sufficiently larger than the maximum value of the error index that can be calculated as an error index related to the representative intra prediction mode. For example, when the number of non-zero transform coefficients is used as the error index, the error index acquisition unit 1104 may set the error amount Z to be twice the number of pixels in the block size. The error index acquisition unit 1104 outputs the error indexes Da and Db of the representative intra prediction modes Pa and Pb to the MPM rank determination unit 1105, respectively.

  The MPM rank determination unit 1105 is based on the error indicator Da related to the representative intra prediction mode Pa of the block A, the error indicator Db related to the representative intra prediction mode Pb of the block B, and a predetermined default MPM input from the error indicator acquisition unit 1104. Thus, the intra prediction mode candidates belonging to the MPM and their ranks are determined. The default MPM is a predetermined intra prediction mode candidate set as an intra prediction mode candidate set when the reference block is invalid, and other intra prediction mode candidates other than the intra prediction mode of the reference block. The MPM rank determination unit 1105 outputs the determined intra prediction mode candidates and information on their ranks to the intra prediction mode information generation unit 1106.

  In the HEVC scheme, any one or a combination of planar prediction, DC (Direct Current) prediction and vertical reference is used as the default MPM. The intra prediction mode candidate set when the reference block is invalid is DC prediction. The reference block is invalid means that there is no representative intra prediction mode available for the reference block. The reference block is invalid, for example, that the reference block does not exist in the target picture, that the reference block has not been encoded by intra prediction, has not been decoded, or a slice to which the reference block belongs. The reference block exists in a different tile from the tile to which the target block belongs, or is encoded using the I_PCM mode as a predetermined special encoding mode. . The I_PCM mode is an encoding mode in which a code indicating a signal value for each pixel is output as it is.

  When the representative intra prediction mode Pa and the representative intra prediction mode Pb are different and the error index Da is equal to or less than the error index Db, the MPM rank determination unit 1105 sets the MPM 0th intra prediction mode candidate as the representative intra prediction mode Pa. The MPM first intra prediction candidate is defined as the representative intra prediction mode Pb. When the representative intra prediction mode Pa and the representative intra prediction mode Pb are different and the error index Da is larger than the error index Db, the MPM rank determination unit 1105 selects the MPM 0th intra prediction mode candidate as the representative intra prediction mode. The prediction mode Pb is determined, and the MPM first intra prediction mode candidate is determined as the representative intra prediction mode Pa.

  When the representative intra prediction mode Pa and the representative intra prediction mode Pb are different, the MPM order determination unit 1105 determines the second intra prediction mode candidate of the MPM using the same procedure as that of the HEVC scheme. That is, when neither the representative intra prediction mode Pa nor the representative intra prediction mode Pb is the planar prediction, the MPM rank determination unit 1105 determines the planar prediction as the second intra prediction mode candidate of the MPM. Furthermore, when neither the representative intra prediction mode Pa nor the representative intra prediction mode Pb is DC prediction, the MPM rank determination unit 1105 determines planar prediction as the second intra prediction mode candidate of the MPM. In other cases, the MPM rank determination unit 1105 determines the second intra prediction candidate of the MPM as the vertical direction reference.

  When the representative intra prediction mode Pa and the representative intra prediction mode Pb are equal, the MPM order determination unit 1105 determines each MPM intra prediction mode candidate using the same procedure as in the HEVC scheme. That is, when the representative intra prediction mode Pa and the representative intra prediction mode Pb are planar prediction or DC prediction, the MPM rank determination unit 1105 determines the MPM 0, 1, and 2 intra prediction mode candidates as the planar prediction, respectively. , DC prediction, vertical reference. When the representative intra prediction mode Pa and the representative intra prediction mode Pb are directional predictions, the MPM rank determining unit 1105 sets the MPM 0, 1, and 2 intra prediction mode candidates as representative intra prediction modes, respectively. The intra prediction mode in the same reference direction as Pa and Pb, the reference directions of the representative intra prediction modes Pa and Pb are adjacent to the counterclockwise intra prediction mode, and the reference directions of the representative intra prediction modes Pa and Pb are adjacent to the clockwise direction. It is determined as an intra prediction mode candidate.

  For invalid reference blocks, the MPM order determination unit 1105 determines the representative intra prediction mode as DC prediction, as in the HEVC scheme. That is, when the block A is invalid, the MPM rank determination unit 1105 determines the DC prediction as the representative intra prediction mode Pa. When the block B is invalid, the MPM rank determination unit 1105 determines DC prediction as the representative intra prediction mode Pb.

  The intra prediction mode information generation unit 1106 generates intra prediction mode information indicating the MPM intra prediction mode candidates input from the MPM rank determination unit 1105 and their ranks. The intra prediction mode information generation unit 1106 outputs the generated intra prediction mode information to the candidate determination unit 1107.

  Intra prediction mode information is input from the intra prediction mode information generation unit 1106 to the candidate determination unit 1107, and an intra prediction image block from the intra prediction image generation unit 1101 and a residual image block from the subtraction unit 1102 for each intra prediction mode. Is entered. The candidate determination unit 1107 calculates an error amount indicating the magnitude of the residual for each intra prediction mode, and determines the intra prediction mode with the smallest calculated error amount as the representative intra prediction mode of the target block. For example, the candidate determination unit 1107 may perform rate-distortion optimization when determining the representative intra prediction mode. In the rate-distortion optimization, an error amount J is a sum of squares or absolute value sums of quantized transform coefficients of a residual image block, and a multiplication value of a bit length of the quantized transform coefficients and a predetermined constant λ. This is a method for determining the minimum intra prediction mode. The constant λ is determined based on the quantization parameter. As the quantization parameter, the quantization width of the orthogonal transform coefficient is used. The candidate determination unit 1107 outputs the predicted image block related to the determined representative intra prediction mode to the subtraction unit 101 via the selection unit 112. Further, the candidate determination unit 1107 stores the index of the target block in association with the determined representative intra prediction mode and error index in a storage unit (not shown) of the intra predictor 1100. The stored intra prediction mode and error index are used as the intra prediction mode and error index of the reference block after the target block is changed.

  The candidate determination unit 1107 determines whether or not the determined representative intra prediction mode is included in the MPM intra prediction mode candidates indicated by the intra prediction mode information input from the intra prediction mode information generation unit 1106. When determining that it is included, the candidate determination unit 1107 identifies the rank of the same intra prediction mode candidate as the representative intra prediction mode. Then, the candidate determination unit 1107 uses the intra prediction flag indicating that the representative intra prediction mode is included in the MPM intra prediction mode candidates and the bit sequence indicating the identified rank number as the representative intra prediction mode information, and has a variable length. The data is output to the encoding unit 104.

  When determining that the representative intra prediction mode is not included in the MPM intra prediction mode candidates, the candidate determination unit 1107 rearranges the mode numbers of the MPM intra prediction mode candidates in ascending order. An integer value of 0 to 34 is set as a mode number for identifying each of the plurality of intra prediction modes. The candidate determination unit 1107 decrements the mode number of the representative intra prediction mode by 1 when the mode number of the representative intra prediction mode is greater than or equal to the mode number of the 0th intra prediction mode candidate of the MPM. The candidate determination unit 1107 further decreases the mode number of the representative intra prediction mode by 1 when the mode number of the reduced representative intra prediction mode is greater than or equal to the mode number of the first intra prediction mode candidate of the MPM. The candidate determination unit 1107 further decreases the mode number of the representative intra prediction mode by 1 when the mode number of the representative intra prediction mode further decreased is equal to or greater than the mode number of the second intra prediction mode candidate of the MPM. Then, the candidate determination unit 1107 determines the mode number of the representative intra prediction mode as the adopted prediction mode number. With these procedures, the adoption prediction mode number is expressed as any value from 0 to 31, so that the amount of information fits in 5 bits. The candidate determination unit 1107 is a variable-length encoding unit 104 using, as representative intra prediction mode information, a bit string indicating an intra prediction flag indicating that the representative intra prediction mode is not included in the MPM intra prediction mode candidates and the adopted prediction mode number. Output to.

  In the HEVC scheme, a truncated rice binarization (TR) scheme is employed in binarization processing when variable-length encoding processing is performed on MPM intra prediction mode candidate numbers. The TR method is one method for binarizing an input value that is any integer from 0 to the maximum value cMax. The TR method is a method of converting a bit string formed by sequentially arranging “1” values at the beginning of the number equal to the input value and adding “0” at the end. However, if the input value is equal to the maximum value cMax, the trailing “0” is omitted. Therefore, when the input value is any value from 0 to cMax-1, the smaller the input value, the smaller the code amount of the bit string generated by the TR method. When the number of MPM intra prediction mode candidates is 3, the maximum value cMax is 2, and therefore the number of intra prediction mode candidates as an input value is represented by one of 0 to 2.

In the conventional HEVC scheme, when the representative intra prediction mode Pa and the representative intra prediction mode Pb are different, the representative intra prediction modes Pa and Pb are determined as the first and second intra prediction mode candidates of the MPM regardless of the error index. It was.
On the other hand, the MPM rank determination unit 1105 according to the present embodiment determines the intra prediction mode candidate having a higher rank (a smaller number value) as the representative intra prediction mode for each reference block having a smaller error index. Therefore, there is a high possibility that the representative intra prediction mode of the target block that gives the predicted image block having the smallest error amount from the input image matches the intra prediction mode candidate with the higher rank. That is, since the number of intra prediction mode candidates that match the representative intra prediction mode is reduced, the code amount after binarization is reduced. The representative intra prediction mode is used in the image decoding apparatus 2 to generate a predicted image in the target block.

(Intra prediction mode)
Next, an example of the intra prediction mode will be described. In the HEVC scheme, there are 35 types of intra prediction modes. Each intra prediction mode has a mode number of 0 to 34. The 35 kinds of intra prediction modes are classified into directionality prediction, DC prediction, and planar prediction. The directionality prediction is an intra that calculates the pixel value of each pixel in the target block by interpolating the pixel values of the pixels closest to the left and right sides of the reference direction line segment among the predetermined reference pixels in the reference block. Prediction mode. As shown in FIG. 5A, the reference directions are 33 directions from the diagonally lower left to the diagonally upper right, and any one of the mode numbers 2 to 34 is assigned in ascending order clockwise in each direction. The closer the reference direction is to the horizontal direction, the denser the distribution is. In the example shown in FIG. 5A, the starting point of the arrow indicating the reference direction is set at the center of the target block.

  DC prediction is an intra prediction mode in which an average value of pixel values between predetermined reference pixels in the reference block is calculated as a pixel value of each pixel in the target block Bk. In the example shown in FIG. 5B, the reference pixels referred in the DC prediction are 2N reference pixels indicated by black circles. N indicates the number of pixels corresponding to the length of one side of the target block. As the reference pixel, a pixel closest to the upper side and the left side of the target block in the reference block is used. However, in a block having a block size of 32 × 32 pixels or less, as an exception process, the pixel value of each pixel in the uppermost row and the leftmost column of the target block is updated with the average value of the pixel value of the nearest reference pixel. The In FIG. 5B, the area Ep filled with shading indicates the uppermost row and the leftmost column of the target block. The mode number for DC prediction is 1.

  Planar prediction is an intra prediction mode in which pixel values of each of four reference pixels are interpolated to calculate the pixel value of each pixel in the target block. As reference pixels, the left (-1, y), the upper (x, -1) of the pixel (x, y) in the target block, the upper right (N, -1), and the lower left (- 1, N) pixels are used. (X, y) and the like indicate the coordinate value of each pixel with the position of the upper left pixel of the target block as the origin. The four reference pixels used for calculation of the pixel value of the target pixel are indicated by black circles surrounded by a dashed-dotted circle in FIG.

(Intra prediction unit)
Next, the configuration of the intra prediction unit 207 of the image decoding device 2 according to the present embodiment will be described. FIG. 6 is a schematic block diagram illustrating a configuration of the intra prediction unit 207 according to the present embodiment. The intra prediction unit 207 includes an intra predictor 2070. The intra predictor 2070 includes an MPM determination unit 2071, an intra prediction mode information generation unit 2074, a candidate determination unit 2075, and an intra prediction image generation unit 2076.

  Similar to the MPM determination unit 1103, the MPM determination unit 2071 determines the MPM and the MPM in-order based on the error index of each decoded reference block and the representative intra prediction mode. The MPM determination unit 2071 includes an error index acquisition unit 2072 and an MPM rank determination unit 2073.

  The error index acquisition unit 2072 acquires the representative intra prediction modes Pa and Pb and the error indexes Da and Db of the reference blocks A and B adjacent to the target block from a storage unit (not shown) included in the intra predictor 2070. . The error index acquisition unit 2072 acquires the same type of error index as the error index acquired by the error index acquisition unit 1104. For the representative intra prediction mode of the invalid reference block, the error index acquisition unit 2072 has a value Z sufficiently larger than the maximum value of the error index that can be calculated as the error index, similarly to the error index acquisition unit 1104. Determine. The error index acquisition unit 2072 outputs the representative intra prediction modes Pa and Pb and the respective error indexes Da and Db to the MPM rank determination unit 2073.

  The MPM rank determining unit 2073 uses the same method as the MPM rank determining unit 1105 to input the error index Da related to the representative intra prediction mode Pa, the error index Db related to the representative intra prediction mode Pb, and a predetermined value input from the error index acquiring unit 2072. Intra prediction mode candidates belonging to the MPM and their rank are determined based on the default MPM. The MPM rank determination unit 2073 outputs the determined intra prediction mode candidates and information on their ranks to the intra prediction mode information generation unit 2074.

  Here, when the representative intra prediction mode Pa and the representative intra prediction mode Pb are different and the error index Da is equal to or less than the error index Db, the MPM order determination unit 2073 represents the representative intra prediction as the MPM 0th intra prediction mode candidate. The mode Pa is set and the representative intra prediction mode Pb is set as the first MPM intra prediction candidate. When the representative intra prediction mode Pa and the representative intra prediction mode Pb are different and the error index Da is larger than the error index Db, the MPM rank determination unit 2073 displays the representative intra as a 0th intra prediction mode candidate of MPM. The prediction mode Pb is set, and the MPM first intra prediction mode candidate is set as the representative intra prediction mode Pa.

  When the representative intra prediction mode Pa and the representative intra prediction mode Pb are different, the MPM order determination unit 2073 determines the second intra prediction mode candidate of the MPM using the same procedure as in the HEVC scheme. Also, when the representative intra prediction mode Pa and the representative intra prediction mode Pb are equal, the MPM order determination unit 2073 determines each intra prediction mode candidate of the MPM using the same procedure as in the HEVC scheme. For invalid reference blocks, the MPM rank determination unit 2073 determines the representative intra prediction mode as DC prediction.

  The intra prediction mode information generation unit 2074 generates MPM intra prediction mode candidates input from the MPM order determination unit 2073 and intra prediction mode information indicating their ranks. The intra prediction mode information generation unit 2074 outputs the generated intra prediction mode information to the candidate determination unit 2075.

  The candidate determination unit 2075 receives the intra prediction flag and the bit number indicating the number of the rank of the candidate of the intra prediction mode or the number of the adopted prediction mode as the representative intra prediction mode information of the target block from the variable length decoding unit 201. In addition, intra prediction mode information is input from the MPM rank determination unit 2073 to the candidate determination unit 2075. The candidate determination unit 2075 determines whether or not the representative intra prediction mode is included in the MPM intra prediction mode candidates based on the intra prediction flag. When determining that it is included, the candidate determination unit 2075 selects an intra prediction mode candidate having a rank indicated by the number included in the representative intra prediction mode information from the MPM indicated by the intra prediction mode information. The candidate determination unit 2075 determines that the selected intra prediction mode candidate is the representative intra prediction mode of the target block.

  When determining that the representative intra prediction mode is not included in the MPM intra prediction mode candidates, the candidate determination unit 2075 rearranges the mode numbers of the MPM intra prediction mode candidates in ascending order. The candidate determination unit 2075 increments the adoption prediction mode number by 1 when the adoption prediction mode number included in the intra prediction mode information is greater than or equal to the mode number of the MPM 0th intra prediction mode candidate. When the increased adoption prediction mode number is equal to or greater than the mode number of the first intra prediction mode candidate of MPM, the candidate determination unit 2075 further increases the adoption prediction mode number by one. The candidate determination unit 2075 further increases the adoption prediction mode number by 1 when the adopted prediction mode number further increased is equal to or greater than the mode number of the second intra prediction mode candidate of the MPM. Then, the candidate determination unit 2075 determines the representative prediction mode of the adoption prediction mode number target block. The candidate determination unit 2075 outputs the information on the determined representative intra prediction mode to the intra predicted image generation unit 2076.

  The intra predicted image generation unit 2076 refers to the decoded image in the reference block, and performs intra prediction using the representative intra prediction mode indicated by the information input from the candidate determination unit 2075. The intra predicted image generation unit 2076 outputs the intra predicted image block generated by the intra prediction to the adding unit 204.

(MPM decision processing)
Next, the MPM determination process according to the present embodiment will be described. FIG. 7 is a flowchart showing MPM determination processing according to the present embodiment. In the following description, the case where the MPM determination unit 1103 is the main subject of the MPM determination process is taken as an example, but the MPM determination unit 2071 also performs the process shown in FIG.

(Step S101) The error index acquisition unit 1104 of the MPM determination unit 1103 determines whether or not the block A is invalid. When it is determined to be invalid (YES in step S101), the process proceeds to step S102. When it determines with it being effective (step S101 NO), the error parameter | index acquisition part 1104 acquires the error parameter | index which concerns on the representative intra prediction mode Pa of the block A, and progresses to the process of step S103.
(Step S102) The MPM order determination unit 1105 determines the representative intra prediction mode Pa of the block A as DC prediction. The error index acquisition unit 1104 determines the error index Da of the block A as the maximum value Z of the error index. Thereafter, the process proceeds to step S103.

(Step S103) The error index acquisition unit 1104 determines whether or not the block B is invalid. When it is determined to be invalid (YES in step S103), the process proceeds to step S104. When it determines with it being effective (step S103 NO), the error parameter | index acquisition part 1104 acquires the error parameter | index which concerns on the representative intra prediction mode Pb of the block B, and progresses to the process of step S105.
(Step S104) The MPM order determination unit 1105 determines the representative intra prediction mode Pb of the block B as DC prediction. The error index acquisition unit 1104 determines the error index Db of the block B as the maximum error index value Z. Thereafter, the process proceeds to step S105.

(Step S105) The MPM order determination unit 1105 determines whether the representative intra prediction mode Pa of the block A and the representative intra prediction mode Pb of the block B are different. When it is determined that they are different (YES in step S105), the process proceeds to step S106. When it is determined that they are not different (NO in step S105), the process proceeds to step S109.

(Step S106) The MPM rank determination unit 1105 determines whether or not the error index Da related to the representative intra prediction mode Pa is equal to or less than the error index Db related to the representative intra prediction mode Pb. When it is determined that the error index Da is equal to or less than the error index Db (YES in step S106), the process proceeds to step S107. When it is determined that the error index Da is larger than the error index Db (NO in step S106), the process proceeds to step S108.

(Step S107) The MPM order determination unit 1105 determines the representative intra prediction mode Pa as the MPM 0th intra prediction mode candidate MPM [0], and sets the representative intra prediction as the MPM first intra prediction mode candidate MPM [1]. Set to mode Pb. The MPM order determination unit 1105 determines the second intra prediction mode candidate MPM [2] of the MPM by performing the same process as in the HEVC scheme. Thereafter, the process of FIG.

(Step S108) The MPM rank determination unit 1105 determines the representative intra prediction mode Pb as the MPM 0th intra prediction mode candidate MPM [0] and the representative intra prediction as the MPM first intra prediction mode candidate MPM [1]. Set to mode Pa. The MPM order determination unit 1105 determines the second intra prediction mode candidate MPM [2] of the MPM by performing the same process as in the HEVC scheme. Thereafter, the process of FIG.

(Step S109) The MPM order determination unit 1105 determines the MPM 0th to 2nd intra prediction mode candidates MPM [0] to MPM [2] by performing the same process as in the HEVC scheme. Thereafter, the process of FIG.

As described above, the image encoding device 1 and the image decoding device 2 according to this embodiment include the intra predictors 1100 and 2070. The intra predictors 1100 and 2070 include error index acquisition units 1104 and 2072 and MPM rank determination units 1105 and 2073. The error index acquisition units 1104 and 2072 acquire the error index of the predicted image obtained by the intra prediction using the representative intra prediction mode of the reference blocks around the target block. When adopting the representative intra prediction mode of the reference block as an intra prediction mode candidate, the MPM rank determination units 1105 and 2073 determine the rank of the intra prediction mode candidate in ascending order of the error index.
According to this configuration, there is a high possibility that the representative intra prediction mode that gives the intra prediction image that minimizes the amount of error from the input image block matches the intra prediction mode candidate with the higher rank. The number of intra prediction mode candidates related to encoding of the representative intra prediction mode, that is, the number of encoded bits can be reduced.

Further, the error index acquisition units 1104 and 2072 acquire the number of bits of a bit string obtained by variable-length encoding the data of the prediction residual quantization transform coefficient based on the intra prediction image as an error index.
According to this configuration, the ranking of the intra prediction mode candidates is determined in ascending order of the number of bits of the bit string of the prediction residual based on the intra prediction image generated using each intra prediction mode candidate. There is a high possibility that an intra prediction mode candidate having a small number of bits as the size of the prediction residual matches the representative intra prediction mode related to the target block. Therefore, it is possible to reduce the number of encoded bits related to encoding in the representative intra prediction mode.

Further, the error index acquisition units 1104 and 2072 acquire the number of non-zero transform coefficients having a value other than zero among the quantized transform coefficients of the prediction residual based on the intra prediction image as the error index.
According to this configuration, the intra prediction mode candidates are ranked with a small number of non-zero transform coefficients of the prediction residual based on the intra prediction image generated using each intra prediction mode candidate. The possibility of matching with the representative intra prediction mode related to the block increases. Therefore, it is possible to reduce the number of encoded bits related to encoding in the representative intra prediction mode.

Further, the error index acquisition units 1104 and 2072 acquire the sum of squares of the quantized transform coefficients of the prediction residual based on the intra prediction image as the error index.
According to this configuration, the intra prediction mode candidates whose ranks of the quantized transform coefficients of the prediction residuals based on the intra prediction images generated using the respective intra prediction mode candidates are small are the ranks of the intra prediction mode candidates. The possibility of matching with the representative intra prediction mode related to the target block increases. Therefore, it is possible to reduce the number of encoded bits related to encoding in the representative intra prediction mode.

Further, the error index acquisition units 1104 and 2072 acquire the absolute value sum of the quantized transform coefficients of the prediction residual based on the intra prediction image as an error index.
According to this configuration, the intra prediction mode candidates having a small absolute value sum of the quantized transform coefficients of the prediction residuals based on the intra prediction images generated using the respective intra prediction mode candidates. There is a high possibility that it matches the representative intra prediction mode related to the target block. Therefore, it is possible to reduce the number of encoded bits related to encoding in the representative intra prediction mode.

(Second Embodiment)
Next, a second embodiment of the present invention will be described with reference to the drawings. About the same structure as the above-mentioned embodiment, the same code | symbol is attached | subjected and the description is used. The intra predictors 1100 and 2070 according to the present embodiment execute an intra prediction mode candidate group determination process that is a more generalized MPM determination process based on the HEVC scheme shown in FIG. The image encoding device 1 and the image decoding device 2 according to the present embodiment according to the present embodiment are configured to include the intra predictors 1100 and 2070, respectively. Note that a group of intra prediction mode candidates is also referred to as MPM in this embodiment.

(MPM decision processing)
Next, the MPM determination process according to the present embodiment will be described. In the example described below, the number of reference blocks is generalized to M. M is an integer of 2 or more. The number of MPMs is represented as N. MPM ranks 0 to N-1 are present. N is an integer larger than M. The representative intra prediction modes of M reference blocks 1 to M are represented as P [1] to P [M]. For example, the reference block related to the representative intra prediction mode used as the MPM is not limited to the block above or to the left of the target block as in the HEVC method, and for example, the upper left block and the upper right block of the target block are also used. May be. The error index for each of the reference blocks 1 to M is represented as D [1] to D [M]. The default MPM is represented as L [1], L [2],. However, the representative intra prediction mode set for an invalid reference block is represented by X.

  FIG. 8 is a flowchart showing MPM determination processing according to the present embodiment. In the following description, a case where the MPM determination unit 1103 is an operation subject is taken as an example, but the MPM determination unit 2071 can also be an operation subject. In the process shown in FIG. 8, a threshold value D [0] is introduced. The threshold value D [0] is an adjustment parameter for adjusting the degree to which the default MPM is selected.

(Step S121) The error index acquisition unit 1104 of the MPM determination unit 1103 sets the intra prediction mode that is the same as each of the representative intra prediction modes P [1] to P [M] of valid reference blocks among the M reference blocks. Such an error index is obtained. Further, the error index acquisition unit 1104 determines the error index as the maximum value Z for each invalid reference block among the M reference blocks. Thereafter, the process proceeds to step S122.

(Step S122) The MPM order determination unit 1105 sorts the M error indexes D [1] to D [M] and the threshold value D [0] in ascending order. The threshold value D [0] obtained by sorting and the indexes of the reference blocks 1 to M are represented as i [0] to i [M]. The indexes i [0] to i [M] take integer values of 1 or more. The list composed of the indexes i [0] to i [M] indicates the rearranged reference blocks and threshold rankings. However, the MPM order determination unit 1105 determines the representative intra prediction mode of invalid reference blocks as the intra prediction mode X, and sets the indexes of those reference blocks to 0. Thereafter, the process proceeds to step S123.

(Step S123) The MPM order determination unit 1105 sets the initial values of the indexes x and y to 0 and 1, respectively. The index x is an index for counting the rank of the representative intra prediction mode of the reference block adopted as the MPM intra prediction mode candidate in ascending order of the error index. The index y is an index for counting the rank of the default MPM employed as an MPM intra prediction mode candidate. Thereafter, the process proceeds to step S124.

(Step S124) The MPM order determination unit 1105 determines whether or not the x-th index i [x] is 0. By this determination, it is determined whether or not the xth reference block is invalid. When it is determined that the value is 0 (YES in step S124), the process proceeds to step S129. When it is determined that it is not 0 (NO in step S124), the process proceeds to step S125.

(Step S125) The MPM order determination unit 1105 determines whether or not the i [x] -th intra prediction mode P [i [x]] is included in the MPM. When it is determined that it is included (step S125 YES), the process proceeds to step S128. When it is determined that it is not included (NO in step S125), the process proceeds to step S126.

(Step S126) The MPM order determination unit 1105 adds the i [x] -th intra prediction mode P [i [x]] to the MPM. The rank in the MPM of the intra prediction mode P [i [x]] to be added is the lowest at that time. Therefore, the rank of the intra prediction mode P [i [x]] added to the MPM for the first time is 0th. Thereafter, the process proceeds to step S127.
(Step S127) The MPM order determination unit 1105 adds 1 to the value of the index x (increment). Thereafter, the process proceeds to step S128.
(Step S128) The MPM order determination unit 1105 determines whether the total number of intra prediction modes included in the MPM is less than N. When it is determined that the number is less than N (YES in step S128), the process proceeds to step S124. When it is determined that the number is greater than or equal to N (NO in step S128), the process illustrated in FIG.

(Step S129) The MPM order determination unit 1105 determines whether or not the y-th default MPM L [y] is included in the MPM. If it is determined that it is included (YES in step S129), the process proceeds to step S128. When it is determined that it is not included (NO in step S129), the process proceeds to step S130.

(Step S130) The MPM order determination unit 1105 adds the y-th default MPM L [y] to the MPM. The rank in the MPM of the default MPM L [y] added is the lowest at that time. Thereafter, the process proceeds to step S131.
(Step S131) The MPM order determination unit 1105 adds 1 to the value of the index y (increment). Thereafter, the process proceeds to step S128.

  Note that, in step S122, the error indices may be equal to each other for representative intra prediction modes of a part of two or more reference blocks out of M. For example, between the reference blocks having the same error index, the error index acquisition unit 1104 determines the rank of each reference block in ascending order of the index of the reference block. The error index acquisition unit 1104 may further acquire other types of error indexes, and may determine the rank for each reference block in ascending order of the acquired error indexes.

  In step S122, the representative intra prediction modes of some of the M reference blocks of two or more (eg, K) may be the same. When the same representative intra prediction modes are non-directional predictions, the error index acquisition unit 1104 does not have to provide a rank for each of the K−1 reference blocks. In that case, the MPM order determination unit 1105 adds K-1 default MPMs to the MPM in the process of step S130. In addition, the same representative intra prediction mode may be directional prediction. In that case, the error index acquisition unit 1104 sets the K-1 intra prediction modes related to the directional prediction in the order from the reference direction closer to the reference direction of the representative intra prediction mode as K-1 intra prediction mode candidates. You may choose. The error index acquisition unit 1104 may also sort the intra prediction modes related to K−1 directional predictions based on the error index. Thereby, the intra prediction mode whose error index is larger than the threshold value D [0] is excluded from the MPM as an intra prediction mode with low reliability.

  According to the processing shown in FIG. 8, the smaller the threshold D [0], the higher the degree to which the default MPM is selected as the MPM intra prediction mode candidate. For example, when the threshold value D [0] is set to Z + 1 as a value larger than the maximum value Z, the representative intra prediction mode of the reference block has priority over the default MPM. When the threshold value D [0] is set to a value within a range that may appear as an error index, the default MPM may be selected in preference to the representative intra prediction mode of the reference block.

  Therefore, the error index acquisition unit 1104 may variably set the threshold value D [0] based on the error index range related to the reference block. For example, the error index acquisition unit 1104 sets the threshold value D [0] to a value within the range of the error index related to each reference block. In this case, the default MPM is selected with priority over the same intra prediction mode as the representative intra prediction mode of the reference block related to the error index larger than the threshold value D [0]. This is because such an intra prediction mode is considered to be less reliable than the default MPM because the error index indicating the magnitude of the prediction residual is high.

  When there is a default MPM that gives an error index smaller than the error index related to any reference block, the error index acquisition unit 1104 sets the threshold D [0] as the error index related to the default MPM and the error related to the reference block. You may set to the value between indicators. In that case, the default MPM related to the error index smaller than the threshold D [0] is selected as the MPM with priority over the representative intra prediction mode of the reference block related to the error index larger than the threshold D [0]. .

In setting the threshold value D [0], the error index acquisition unit 1104 may use the number of bits of a bit string obtained by variable length encoding a code indicating a quantized orthogonal transform coefficient as an error index. Therefore, an MPM for encoding the intra prediction mode with a smaller code amount is selected.
The error index acquisition unit 1104 does not necessarily have to change the threshold value D [0] for each picture. The set period of the threshold value D [0] may be a longer period, for example, a predetermined number of pictures, an access unit, or a sequence. Therefore, the processing amount is reduced as compared with the case where the threshold value D [0] is changed for each picture.
The error index acquisition unit 1104 outputs the set threshold value D [0] to the variable length encoding unit 104.

The variable length coding unit 104 includes a threshold value D [0] in the header of the bit stream forming the coded data (for example, HEVC SPS [Sequence Parameter Set] header, PPS [Picture Parameter Set] header, etc.).
The variable length decoding unit 201 extracts the threshold value D [0] from the header of the bit stream that forms the input encoded data, and outputs the extracted threshold value D [0] to the error index acquisition unit 2072. The error index acquisition unit 2072 acquires the threshold value D [0] input from the variable length decoding unit 201, and the error index acquisition unit 2072 and the MPM rank determination unit 2073 receive the threshold value [0] input from the variable length decoding unit 201. ] Is used in the process of FIG.

  In the process shown in FIG. 8, M = 2, N = 3, threshold D [0] = Z + 1, intra prediction mode X related to an invalid reference block is DC prediction, default MPM L [1], L [2] , L [3] are planar prediction, DC prediction, and vertical reference, respectively, the processing is the same as that shown in FIG. 7 based on the HEVC scheme.

As described above, in the intra predictors 1100 and 2070 according to the present embodiment, the MPM rank determination units 1105 and 2073 give reference error blocks that are larger than a predetermined error index threshold as intra prediction mode candidates. The rank of the representative intra prediction mode is made lower than the rank of a predetermined default MPM.
With this configuration, the rank of the intra prediction mode candidate of the representative intra prediction mode related to the reference pixel that gives an error index larger than the predetermined error index threshold is lower than the predetermined default MPM. Therefore, in encoding of the representative intra prediction mode, a default MPM having a smaller error index is preferentially used as an intra prediction mode candidate. Therefore, it is possible to reduce the number of intra prediction mode candidates used in encoding the representative intra prediction mode related to the target block, that is, the number of encoded bits, rather than only the representative intra prediction mode related to the reference block.

Further, the error index acquisition units 1104 and 2072 determine a threshold value of the error index based on the range of the error index related to each intra prediction mode candidate.
With this configuration, in encoding of the representative intra prediction mode, it is determined whether or not the default MPM having a smaller error index than the representative intra prediction mode related to the reference pixel is used preferentially based on a predetermined threshold value of the error index. Is done. Therefore, the number of encoded bits in the representative intra prediction mode is less than that of the reference block representing the representative intra prediction mode that always gives an error index larger than a certain error index threshold than the predetermined default MPM order. can do.

  As described above, the embodiment of the present invention has been described in detail with reference to the drawings. However, the specific configuration is not limited to the above, and various design changes and the like can be made without departing from the scope of the present invention. It is possible to

  For example, the image encoding device 1, the image decoding device 2, and the intra predictor according to the above-described embodiments may each be implemented alone, or the image encoding device 1 and the image decoding device 2 including the intra predictor, May be implemented as an image processing system.

Note that a part of the image encoding device 1 or the image decoding device 2 described above and a part of the intra predictors configured as the intra predictors 1100 and 2070 may be realized by a computer. In that case, the program for realizing the control function may be recorded on a computer-readable recording medium, and the program recorded on the recording medium may be read by the computer system and executed. Here, the “computer system” is a computer system built in the image encoding device 1, the image decoding device 2, or the intra predictors 1100 and 2070, and includes hardware such as an OS and peripheral devices. . The “computer-readable recording medium” refers to a storage device such as a flexible medium, a magneto-optical disk, a portable medium such as a ROM and a CD-ROM, and a hard disk incorporated in a computer system. Furthermore, the “computer-readable recording medium” is a medium that dynamically holds a program for a short time, such as a communication line when transmitting a program via a network such as the Internet or a communication line such as a telephone line, In this case, a volatile memory inside a computer system that serves as a server or a client may be included that holds a program for a certain period of time. The program may be a program for realizing a part of the functions described above, and may be a program capable of realizing the functions described above in combination with a program already recorded in a computer system.
Moreover, you may implement | achieve part or all of the image coding apparatus 1, the image decoding apparatus 2, or the intra predictors 1100 and 2070 in the above-described embodiments as an integrated circuit such as an LSI (Large Scale Integration). Each functional block of the image encoding device 1, the image decoding device 2, or the intra predictors 1100 and 2070 may be individually made into a processor, or a part or all of them may be integrated into a processor. Further, the method of circuit integration is not limited to LSI, and may be realized by a dedicated circuit or a general-purpose processor. In addition, when an integrated circuit technology that replaces LSI appears due to the advancement of semiconductor technology, an integrated circuit based on the technology may be used.

DESCRIPTION OF SYMBOLS 1 ... Image coding apparatus, 2 ... Image decoding apparatus, 101 ... Subtraction part, 102 ... Orthogonal transformation part, 103 ... Quantization part, 104 ... Variable length coding part, 105 ... Dequantization part, 106 ... Inverse orthogonal transformation 107: Adder, 108 ... Loop filter, 109 ... Reference memory, 110 ... Intra prediction unit, 111 ... Inter prediction unit, 112 ... Selection unit, 201 ... Variable length decoding unit, 202 ... Inverse quantization unit, 203 ... Inverse orthogonal transform unit, 204 ... addition unit, 205 ... loop filter, 206 ... reference memory, 207 ... intra prediction unit, 208 ... inter prediction unit, 1100 ... intra prediction unit, 1101 ... intra prediction image generation unit, 1102 ... subtraction unit DESCRIPTION OF SYMBOLS 1103 ... MPM determination part, 1104 ... Error parameter | index acquisition part, 1105 ... MPM ranking determination part, 1106 ... Intra prediction mode information generation part, 1107 ... Candidate judgment Department, 2070 ... intra-predictor, 2071 ... MPM determining unit, 2072 ... error indicator acquiring unit, 2073 ... MPM order determination unit, 2074 ... intra-prediction mode information generation unit, 2075 ... candidate determination unit, 2076 ... intra-prediction image generation unit

Claims (10)

An error index acquisition unit for acquiring an error index between a predicted image and an input image obtained using an intra prediction mode determined for reference blocks around the target block;
When using the intra prediction mode defined for the reference block as the intra prediction mode candidate of the target block, a rank determining unit that determines the rank of the intra prediction mode candidates in ascending order of the error index;
An intra predictor comprising: The rank determining unit lowers the rank of the intra prediction mode of the reference block that gives the error index larger than a predetermined error index threshold as the intra prediction mode candidate lower than the rank of the predetermined intra prediction mode candidate. The intra predictor according to claim 1, wherein: The intra predictor according to claim 2, wherein the error index acquisition unit determines a threshold of the error index based on an error index range related to an intra prediction mode determined for the reference block. The error index acquisition unit acquires the number of bits of a bit string obtained by variable-length encoding a quantized transform coefficient of a prediction residual based on the prediction image as the error index. 4. The intra predictor according to any one of 3. The error index acquisition unit acquires the number of non-zero transform coefficients having a non-zero value among the quantized transform coefficients of the prediction residual based on the predicted image as the error index. The intra predictor according to claim 3.   The intra predictor according to any one of claims 1 to 3, wherein the error index acquisition unit acquires a square sum of quantized transform coefficients of a prediction residual based on the predicted image as the error index.   The intra predictor according to any one of claims 1 to 3, wherein the error index acquisition unit acquires an absolute value sum of quantization transform coefficients of a prediction residual based on the predicted image as the error index. The intra predictor according to any one of claims 1 to 7,
Based on the error amount between the prediction image obtained by performing intra prediction using each predetermined intra prediction mode with reference to the decoded image of the reference block and the input image of the target block, one of the intra prediction modes is selected. Selected,
A candidate determination unit that determines the rank of the same intra prediction mode candidate as the selected intra prediction mode;
An image encoding device comprising: An intra predictor according to any one of claims 1 to 7,
An image decoding apparatus, comprising: generating a decoded image of the target block using a prediction image obtained using the same intra prediction mode as the intra prediction mode candidate of the input rank. An error index acquisition unit for acquiring an error index between a predicted image and an input image obtained using an intra prediction mode determined for reference blocks around the target block in a computer;
When using the intra prediction mode defined for the reference block as the intra prediction mode candidate of the target block, a rank determining unit that determines the rank of the intra prediction mode candidates in ascending order of the error index;
A program for causing a computer to function as an intra predictor.

Images