JP6616242B2 - Video coding apparatus, block structure determination method, and computer program - Google Patents

Description

  The present invention relates to a video encoding device, a block structure determination method, and a computer program.

  H.264 is a standard for video coding. H.264 / AVC (Advanced Video Coding) and H.264. A video encoding method called H.265 / HEVC (High Efficiency Video Coding) has been formulated. While HEVC achieves a high compression ratio, it is known that the amount of calculation of encoding processing is larger than that of the conventional standard. One reason for this is that the variation of coding blocks in the screen has increased. In HEVC, an image region to be encoded is divided into square block units called LCUs (Largest Coding Units), and the LCUs are recursively up to 3 times in block units called CUs (Coding Units). Can be divided into 4 equal parts.

  The LCU can be set to a maximum of 64 × 64 pixels. In that case, the CU can be set to 64 × 64 pixels, 32 × 32 pixels, 16 × 16 pixels, and 8 × 8 pixels. Combining CU (maximum 4 patterns) blocks for each LCU and determining an optimal block structure enables efficient compression. However, since the number of combinations becomes enormous, if all combinations are performed, there is a problem that the calculation load of encoding increases. Therefore, a method for selecting an optimal block structure with a small amount of calculation is desired.

  For example, a method of restricting block size candidates using the minimum CU size of adjacent blocks has been proposed (see, for example, Patent Document 1). In the technique described in Patent Document 1 (hereinafter referred to as “Prior Art 1”), the structure of adjacent blocks needs to be selected in advance, and parallel processing that simultaneously selects the structure of a plurality of blocks is performed. Is difficult to apply. As a technique for solving this problem, a technique for acquiring adjacent blocks from encoded frames has been proposed (see, for example, Non-Patent Document 1). In the technique described in Non-Patent Document 1 (hereinafter referred to as “conventional technique 2”), the maximum division depth of the encoding target block is selected based on the maximum division depth of the reference block of the encoded frame.

Japanese Patent No. 5719401

Guilherme Correa, Pedro Assuncao, Luciano Agostini, Luis A. da Silva Cruz, “Complexity Control of HEVC through Quadtree Depth Estimation”, EuroCon 2013, 1-4 July 2013, Zagreb, Croatia, pp. 81-86

  However, in the conventional technique 2, if there is a block with the largest division even at one location, the blocks of all patterns are calculated, and there is a case where the amount of calculation cannot be reduced. For this reason, a method of determining for each block size (hereinafter referred to as “conventional technology 2 ′”) by combining the conventional technology 2 with the conventional technology 1 and the like as shown in FIG. 21 is also conceivable. However, since the temporal reference block does not necessarily have a high correlation with the encoding target block, there is a problem that if the block size is individually determined when the correlation is low, the encoding efficiency decreases.

  The present invention has been made to solve the above problems, and an object of the present invention is to provide a video encoding apparatus, a block structure determination method, and a computer program that reduce the amount of calculation of block structure determination processing while suppressing a decrease in encoding efficiency. Is to provide.

One embodiment of the present invention is configured with one block size generated by dividing an image into a plurality of regions and recursively dividing the region into blocks, or a combination of blocks having a plurality of block sizes. A block structure determining unit that determines a block structure to be processed, and an encoding unit that performs encoding based on the determined block structure, and the block structure determining unit includes, for each block size, the image of the image A block dividing unit that divides an area into blocks of a predetermined size, the size of the divided block, and the size of the block to which the center of the area referred to by the block belongs are equal or the size of the block to which the center belongs is If one step higher, each of said blocks, a first candidate block determination unit determines that the first candidate block, before If the number of block groups obtained by dividing one step larger size of the blocks than a predetermined size to the predetermined size is determined as the first candidate block is not less than a predetermined value, belonging to the blocks A second candidate block determination unit that determines that all blocks are second candidate blocks; a block size evaluation value calculation unit that calculates an evaluation value for the block determined to be the second candidate block; and the evaluation And a final block selection unit that selects a combination of the blocks that divide the region of the image based on the value.

One aspect of the present invention is the video encoding device, wherein the block division unit divides the image into a plurality of predetermined sizes, the first candidate block determination unit, and the second The candidate block determination unit performs the determination for each of the plurality of sizes, and when the predetermined region of the image is determined to be duplicated by the second candidate block determination unit as the second candidate block, The image processing apparatus further includes an overlapping area processing unit that excludes any block for a predetermined area of the image from the second candidate block .

One aspect of the present invention is the video encoding device described above, wherein the first candidate block determination unit corresponds to the divided block and the region of the image for each prediction mode applied to the image. And determining whether each of the blocks is a first candidate block based on a reference block that divides a region in a reference picture to be processed, and the second candidate block determination unit determines whether the block is a predetermined size or not. for one-step larger size block divided and obtained block group to the predetermined size, for each of the prediction mode, and the ratio of the first candidate block included in the block group, the prediction mode and the block Whether or not all the blocks in the block group are to be the second candidate blocks is determined based on the likelihood determined in advance according to the size .

  One aspect of the present invention is the video encoding device described above, in which the first candidate block determination unit determines whether the block size of each of the divided blocks for any of the prediction modes is the reference block. When the block size is the same as the block size or the block size that is one step smaller than the block size of the reference block, the first candidate block is used, and each of the divided blocks is divided into prediction modes other than the prediction mode. When the block size is the same as the block size of the reference block or the same block size as the block size of the reference block, the block is determined as the first candidate block.

One aspect of the present invention is the above-described video encoding apparatus, different likelihood for each combination of the prediction modes and the block size is predetermined, the second candidate block determination unit of the block group Based on the block size and the prediction mode, the likelihood of the block group for each prediction mode is calculated, and based on the total value of the calculated likelihoods, all the blocks in the block group are second candidate blocks. It is determined whether or not.

One aspect of the present invention is the video encoding device described above, wherein the second candidate block determination unit determines that all blocks in the block group are second candidate blocks. When adding a block that is not the first candidate block as the second candidate block, the block has a large total likelihood value among the total likelihood values obtained in each of a plurality of prediction modes. The second candidate block corresponding to the prediction mode is added.

One embodiment of the present invention is configured by one block size generated by dividing an image region into blocks and recursively dividing the region into blocks, or a combination of blocks having a plurality of block sizes. A block structure determining step for determining a block structure to be performed, and an encoding step for performing encoding based on the determined block structure. In the block structure determining step, for each block size, the image A block dividing step for dividing the area into blocks of a predetermined size, the size of the divided block, and the size of the block to which the center of the area to which the block refers belongs, or the size of the block to which the center belongs determine if There one step higher, each of said blocks, when the first candidate block To a first candidate block determination step, said number of predetermined one stage larger size block group block obtained by dividing the predetermined size from the size is determined as the first candidate block is given for more than a value, calculates an evaluation value of all the blocks belonging to the block group and a second candidate block determination step determines to a second candidate block for the second candidate block and determined to be the block A block structure determination method that performs a block size evaluation value calculation step and a final block selection step that selects a combination of the blocks that divide the image area based on the evaluation value.

One embodiment of the present invention is configured by one block size generated by dividing an image region into blocks and recursively dividing the region into blocks, or a combination of blocks having a plurality of block sizes. A block structure determination step for determining a block structure to be performed, and an encoding step for performing encoding based on the determined block structure, and causing the computer to execute the block structure determination step for each block size. A block dividing step for dividing the image area into blocks of a predetermined size, and the size of the divided block is equal to the size of the block to which the center of the area referred to by the block belongs, or to which the center belongs If the size of one step higher, each of said blocks, a first candidate A first candidate block determination step of determining that a lock, the predetermined one stage larger size block group block obtained by dividing the predetermined size from the size is determined as the first candidate block If the number there are greater than a predetermined value, and a second candidate block determination step determines to all blocks belonging to the block group to a second candidate block for the second candidate block and determined to be the block A computer program for executing a block size evaluation value calculation step for calculating an evaluation value and a final block selection step for selecting a combination of the blocks for dividing the region of the image based on the evaluation value.

  According to the present invention, it is possible to reduce the calculation amount of the block structure determination process while suppressing a decrease in encoding efficiency.

It is a block diagram which shows the structure of the video coding apparatus in 1st embodiment of this invention. It is a block diagram which shows the structure of the block structure determination part in 1st embodiment. It is a figure which shows the relationship between the encoding object LCU and the reference block in a reference picture in 1st embodiment. It is a figure which shows the relationship between the encoding object block and reference block in 1st embodiment. It is a flowchart which shows the block structure determination process in 1st embodiment. It is a figure which shows the process by the 1st candidate block determination part in 1st embodiment. It is a figure which shows the process by the 2nd candidate block determination part in 1st embodiment. It is a figure which shows the process by the block size evaluation value calculation part and last block selection part in 1st embodiment. It is a block diagram which shows the structure of the block structure determination part in 2nd embodiment of this invention. It is a flowchart which shows the block structure determination process in 2nd embodiment. It is a figure which shows the process by the duplication area | region process part in 2nd embodiment. It is a figure which shows the process by the 2nd candidate block determination part in 2nd embodiment. It is a figure which shows the process by the block size evaluation value calculation part and last block selection part in 2nd embodiment. It is a block diagram which shows the structure of the block structure determination part in 3rd embodiment of this invention. It is a figure which shows the structure of the likelihood table in 3rd embodiment. It is a flowchart which shows the block structure determination process in 3rd embodiment. It is a figure which shows the process by the 1st candidate block determination part in 3rd embodiment. It is a figure which shows the process (the 1) by the 2nd candidate block determination part in 3rd embodiment. It is a figure which shows the process (the 2) by the 2nd candidate block determination part in 3rd embodiment. It is a figure which shows the process by the block size evaluation value calculation part and last block selection part in 3rd embodiment. It is a figure which shows an example of the conventional block structure determination part.

(First embodiment)
Hereinafter, embodiments of the present invention will be described in order from the first embodiment with reference to the drawings. FIG. 1 is a block diagram showing a configuration of a video encoding device 1 according to the first embodiment of the present invention. The video encoding device 1 includes a block structure determination unit 10, a subtractor 11, an orthogonal transformation / quantization unit 12, a variable length coding unit 13, an inverse quantization / inverse orthogonal transformation unit 14, an adder 15, and an intra prediction unit 16. , An intra / inter switch 17, a loop filter unit 18, a decoded picture memory unit 19, and an inter prediction unit 20.

  The video encoding device 1 receives an encoding target picture constituting an encoding target video from an external device or the like for each LCU, and outputs encoded data corresponding to the block of the LCU to another external device or the like. . The encoding target picture is raster-scanned in LCU size units by an external device or the like, input to the video encoding device 1, and the video encoding device 1 repeatedly performs processing in the order of the raster scan. The encoding target picture is encoded. In the video encoding device 1, the block structure determination unit 10 determines the block structure of the LCU input from an external device or the like, that is, selects an appropriate combination of blocks. Further, the block structure determining unit 10 recursively divides the LCU into blocks according to the determined block structure, generates a CU, repeats in a predetermined order, and outputs the generated CU to the subtractor 11. .

  The subtractor 11 subtracts the predicted image data output from either the intra prediction unit 16 or the inter prediction unit 20 from the CU image data, and outputs the subtracted difference to the orthogonal transform / quantization unit 12. The orthogonal transform / quantization unit 12 performs orthogonal transform and quantization on the difference image data output from the subtractor 11, and outputs the result to the variable length coding unit 13 and the inverse quantization / inverse orthogonal transform unit 14. To do. The variable length encoding unit 13 performs variable length encoding on the quantization coefficient output from the orthogonal transform / quantization unit 12 and outputs the result as encoded data. The inverse quantization / inverse orthogonal transform unit 14 performs inverse quantization and inverse orthogonal transform on the quantization coefficient output from the orthogonal transform / quantization unit 12 and outputs the result to the adder 15. The adder 15 adds the predicted image data output from either the intra prediction unit 16 or the inter prediction unit 20 to the image data output from the inverse quantization / inverse orthogonal transform unit 14, loops, Output to the filter unit 18.

  The intra prediction unit 16 generates intra prediction image data of the block to be encoded, using the image data output from the adder 15 after adding the prediction image data to the image data as reference image data. The loop filter unit 18 applies a loop filter to the image data output from the adder 15 and outputs the filtered image data to the decoded picture memory unit 19. The decoded picture memory unit 19 stores image data to which a loop filter is applied. The inter prediction unit 20 uses the image data stored in the decoded picture memory unit 19 as reference image data, and generates an inter prediction image of the encoding target block. The intra / inter changeover switch 17 switches between the intra prediction unit 16 and the inter prediction unit 20 according to the prediction mode of the encoding target block, and subtracts 11 and an adder for predicted image data corresponding to each prediction mode. 15 and output.

  FIG. 2 is a block diagram illustrating an internal configuration of the block structure determination unit 10. The block structure determination unit 10 includes a block division unit 101, a first candidate block determination unit 102, a second candidate block determination unit 103, a block size evaluation value calculation unit 104, and a final block selection unit 105. The block division unit 101 divides the LCU of the encoding target picture into blocks of 32 × 32 pixels, 16 × 16 pixels, and 8 × 8 pixels. The block dividing unit 101 is divided into an LCU having a maximum block size of 64 × 64 pixels, an LCU divided into 32 × 32 pixels, an LCU divided into 16 × 16 pixels, and an 8 × 8 pixel. The LCU is output to the first candidate block determination unit 102.

  The first candidate block determination unit 102 includes a 64 × 64 block determination unit 1021, a 32 × 32 block determination unit 1022, a 16 × 16 block determination unit 1023, and an 8 × 8 block determination unit 1024. Each determination unit of the first candidate block determination unit 102 (that is, a 64 × 64 block determination unit 1021, a 32 × 32 block determination unit 1022, a 16 × 16 block determination unit 1023, and an 8 × 8 block determination unit 1024) The LCU corresponding to each block size output from the dividing unit 101 is received. That is, the 64 × 64 block determination unit 1021 receives an LCU of 64 × 64 pixels, the 32 × 32 block determination unit 1022 receives an LCU divided into 32 × 32 pixels, and the 16 × 16 block determination unit 1023 The 8 × 8 block determination unit 1024 receives the LCU divided into 16 × 16 pixels, and receives the LCU divided into 8 × 8 pixels. The 64 × 64 block determination unit 1021, the 32 × 32 block determination unit 1022, the 16 × 16 block determination unit 1023, and the 8 × 8 block determination unit 1024 determine whether each block of the LCU corresponds to the first candidate block. judge.

  Here, the determination of whether or not the first candidate block determination unit 102 corresponds to the first candidate block is performed as follows. In the case of the 64 × 64 block determination unit 1021, one encoding target block, in the case of the 32 × 32 block determination unit 1022, in the case of four encoding target blocks, and in the case of the 16 × 16 block determination unit 1023, 16 In the case of the 8 encoding target blocks and the 8 × 8 block determination unit 1024, 64 encoding target blocks are to be determined. For each of these encoding target blocks, it is first determined whether or not the following condition (1) is satisfied.

  Condition (1): It is determined that the condition (1) is satisfied when the block size of the block to be encoded is the same as the block size of the reference block or the block size that is one step smaller. Here, as shown in FIG. 3, the reference block is a CU including the center coordinates of the encoding target block in the reference picture indicated by the LCU motion vector. For example, in FIG. 4, when the encoding target block has a size of 64 × 64 pixels, the CU in the reference picture in which the center coordinates exist is the reference block of the upper right 64 × 64 pixels.

  Each determination unit of the first candidate block determination unit 102 outputs the determination result to the second candidate block determination unit 103. The second candidate block determination unit 103 receives the determination result output from the first candidate block determination unit 102 and determines whether or not the following condition (2) is satisfied.

  Condition (2): When the encoding target block has the maximum block size (64 × 64 pixels in the present embodiment), the block that satisfies the condition (1) is determined to satisfy the condition (2) and the second Candidate block. If the block size is not the maximum, the block is divided into four blocks having a block size that is one step larger than the block size. For example, if the block size is 16 × 16 pixels, one of the 32 × 32 pixels that are one step larger is divided into four. A block group belonging to the four 16 × 16 pixel regions is selected. In the block group, when the number of blocks satisfying the condition (1) is equal to or greater than a predetermined number, it is determined that the condition (2) is satisfied, and all the block groups are set as second candidate blocks. When the number of blocks satisfying the condition (1) is less than the predetermined number, it is determined that the condition (2) is not satisfied, and all the block groups are not set as the second candidate blocks. However, an area that has never become the second candidate block in the encoding target LCU is preferentially set as the second candidate block from a small block size so as to become the second candidate block at least once. The predetermined number may be changed for each block size. In this embodiment, “2” is used for a block size of 32 × 32 pixels or more, and “2” is used for a block size of 16 × 16 pixels or less. 3 ”.

  The block size evaluation value calculation unit 104 calculates the evaluation value (hereinafter referred to as evaluation cost) of the block that has become the second candidate block, that is, the RD cost to be calculated based on the following equation (1).

  RD cost = D + λR (1)

  In Expression (1), D represents an error when the encoding target block is encoded, R represents a code amount, and λ represents a Lagrangian constant. Further, the block size evaluation value calculation unit 104 calculates the RD cost for each prediction mode, that is, for each prediction mode of the intra prediction mode by the intra prediction unit 16 and the inter prediction mode by the inter prediction unit 20, and minimizes the RD cost. Let cost be the block evaluation cost.

  The final block selection unit 105 selects, as a block structure, a combination of blocks in which the block size evaluation value calculation unit 104 has the smallest evaluation cost. For example, when comparing the small block size and the large block size in the same block region, the final block selecting unit 105 compares the total evaluation cost of the small block size with the evaluation cost of the large block size, and the evaluation cost is small. A block having a block size is selected as a block constituting the block structure.

  Next, block structure determination processing by the block structure determination unit 10 of the video encoding device 1 according to the first embodiment will be described with reference to FIGS. 5 to 8. FIG. 5 is a flowchart showing the flow of the block structure determination process. The block division unit 101 of the block structure determination unit 10 of the video encoding device 1 converts the LCU of the encoding target picture received from an external device or the like into block sizes of 32 × 32 pixels, 16 × 16 pixels, and 8 × 8 pixels. To divide. The block dividing unit 101 includes an LCU having a maximum block size of 64 × 64 pixels, an LCU divided into 32 × 32 pixels, an LCU divided into 16 × 16 pixels, and an LCU divided into 8 × 8 pixels. Is output to the corresponding determination unit of the first candidate block determination unit 102 (step Sa1).

  The first candidate block determination unit 102 determines whether or not the encoding target block is a first candidate block that satisfies the condition (1), and performs a process of selecting the first candidate block (step Sa2). The processing of step Sa2 is performed by each determination unit in the order of 64 × 64 block determination unit 1021, 32 × 32 block determination unit 1022, 16 × 16 block determination unit 1023, 8 × 8 block determination unit 1024, for example. This is performed for the LCU having the corresponding block size (loop La1). In addition, each determination unit of the first candidate block determination unit 102 repeatedly performs the process of step Sa2 for each of the blocks included in the encoding target block (loop La2). For example, in the case of the 8 × 8 block determination unit 1024, Step Sa2 is repeated 64 times for each of the 64 blocks. When the determining unit of the first candidate block determining unit 102 performs the process of step Sa2 for each of the blocks included in the encoding target block, the loop La2 ends (loop La3). Further, when the determination unit of the first candidate block determination unit 102 performs the processing of step Sa2 on the LCUs having the corresponding block size, the loop La1 is terminated (loop La4).

  FIG. 6 is a diagram illustrating an example of the process of step Sa2 by the first candidate block determination unit 102. In FIG. 6, “◯” indicates that the block satisfies the condition (1), and “X” indicates that the block does not satisfy the condition (1). For example, when the relationship between each block of the encoding target LCU and the reference block is the relationship illustrated in FIG. 4, the 64 × 64 block determination unit 1021 receives the LCU of 64 × 64 pixels output from the block division unit 101. Thus, it is determined in which reference block of the reference picture the center coordinates of the block of 64 × 64 pixels are present. The center coordinates of 64 × 64 pixels are present in the reference block of 64 × 64 pixels at the upper right of the reference picture and have the same block size as the reference block, so the condition (1) is satisfied. Therefore, the 64 × 64 block determination unit 1021 outputs a determination result of “◯”, that is, the first candidate block, as shown in FIG.

  The 32 × 32 block determination unit 1022 receives the LCU divided into 32 × 32 pixels from the block division unit 101, and the center coordinates of each of the four blocks of 32 × 32 pixels are which reference block of the reference picture. To determine if it exists. The upper right block exists in a reference block of 64 × 64 pixels of the reference picture, and corresponds to a block having a block size that is one step lower than the block size of the reference block. Therefore, the upper right block satisfies the condition (1). Similarly, since the upper left block is present in a 32 × 32 pixel block of the reference picture, it corresponds to a block having the same block size as the reference block, and therefore satisfies the condition (1). On the other hand, the lower right block and the lower left block exist in the 16 × 16 pixel block of the reference picture, and therefore the condition (1) is not satisfied. Therefore, the 32 × 32 block determination unit 1022 indicates that the upper right and upper left blocks correspond to the first candidate block as shown in FIG. 6B, that is, the first candidate block, and the lower right and lower left blocks indicate “×”. That is, a determination result that does not correspond to the first candidate block is output.

  Similarly, the 16 × 16 block determination unit 1023 receives the LCU divided into 16 × 16 pixels from the block division unit 101, and the center coordinates of the 16 blocks of 16 × 16 pixels indicate which reference of the reference picture. Determine if it exists in the block. Condition (1) is satisfied when the center coordinates are present in a 16 × 16 pixel block or a reference block of 32 × 32 pixels, and condition (1) is not satisfied when the coordinates are present in a reference block of 64 × 64 pixels. The 16 × 16 block determination unit 1023 outputs a determination result as shown in FIG. Since the 8 × 8 block determination unit 1024 does not satisfy the condition (1) for all 8 × 8 pixels, “×” illustrated in FIG. 6D, that is, determination that all do not correspond to the first candidate block. Output the result.

  Returning to FIG. 5, the second candidate block determination unit 103 receives the determination result by each determination unit of the first candidate block determination unit 102 and determines whether or not the encoding target block satisfies the condition (2). (Step Sa3). FIG. 7 is a diagram illustrating a result of determination by the second candidate block determination unit 103. In FIG. 7, “☆” indicates a block determined as the second candidate block. In FIG. 7, the meaning of “†” means that the block that was not the first candidate block in FIG. 6 is the second candidate block, or the block that was the first candidate block is the first candidate block. This is a symbol indicating a case where the candidate block is not set to two candidate blocks.

  In response to the determination result of the first candidate block of FIG. 6, for the LCU of 64 × 64 pixels, the second candidate block determination unit 103 corresponds to the maximum block size, and further, the LCU of 64 × 64 pixels is Since it also corresponds to the first candidate block, it is determined that the condition (2) is satisfied. Therefore, the second candidate block determination unit 103 outputs “☆”, that is, a determination result corresponding to the second candidate block, as shown in FIG.

  For the LCU divided into 32 × 32 pixels, the second candidate block determination unit 103 is not the maximum block size, and therefore becomes the first candidate block among the blocks belonging to the recursively divided into four areas. Detect the number of blocks that are present. As shown in FIG. 6B, since there are two first candidate blocks for the LCU of 32 × 32 pixels, the second candidate block determination unit 103 detects the number as “2”. For the 32 × 32 pixels, the predetermined number set in advance is “2”, so that the second candidate block determination unit 103 satisfies the condition (2) for the LCU divided into 32 × 32 pixels. A determination result is output, with all blocks existing in the four-divided area as second candidate blocks. Therefore, as shown in FIG. 7B, the second candidate block determination unit 103 outputs all blocks of the LCU of 32 × 32 pixels as “☆”, that is, a determination result corresponding to the second candidate block. .

  For the LCU divided into 16 × 16 pixels, since the second candidate block determination unit 103 is not the maximum block size, for each block belonging to the region further recursively divided into four from the 32 × 32 pixel block, The number of blocks that are first candidate blocks is detected. As shown in FIG. 6C, in the 16 × 16 pixel LCU, since there are three first candidate blocks in the lower left quadrant block, the second candidate block determination unit 103 determines the number of blocks. Is detected as “3”. In addition, since there are two first candidate blocks in the lower right quadrant block, the second candidate block determination unit 103 detects the number as “2”. In addition, since there are two first candidate blocks in the upper left divided four blocks, the second candidate block determination unit 103 detects the number as “2”. Since the first candidate block does not exist in the upper right four-divided block, the second candidate block determination unit 103 detects the number as “0”. For the 16 × 16 pixel block, since the predetermined number is “3”, the second candidate block determination unit 103 satisfies the condition (2) for the lower left quadrant, About other than that, it determines with not satisfy | filling conditions (2). Therefore, the second candidate block determination unit 103 corresponds to “☆”, that is, the second candidate block for blocks belonging to the lower left quadrant area that satisfies the condition (2), and other blocks. , “×”, that is, a determination result not corresponding to the second candidate block is output.

  The second candidate block determination unit 103 determines that all the LCUs divided into 8 × 8 pixels do not satisfy the condition (2) because there is no block that is the first candidate block.

  Next, the block size evaluation value calculation unit 104 calculates an evaluation cost, that is, an RD cost, for each prediction mode for each block that has become the second candidate block based on the above equation (1) (step Sa4). ). The numerical value in each block shown in FIG. 8A is the evaluation cost value calculated by the block size evaluation value calculation unit 104. The final block selection unit 105 selects a combination that minimizes the evaluation cost calculated by the block size evaluation value calculation unit 104 as a block constituting the block structure (step Sa5). As shown in FIG. 8A, first, the final block selection unit 105 evaluates “30”, which is the evaluation cost of the lower left block of 32 × 32 pixels, and the lower left quadrant area of 16 × 16 pixels. The total cost is compared with “25 = 10 + 5 + 5 + 5”. Since the total evaluation cost of the 16 × 16 pixel four-divided area is smaller, the final block selection unit 105 selects the block divided into 16 × 16 pixels.

  Next, the final block selection unit 105 compares “100”, which is the evaluation cost of 64 × 64 pixels, with the total evaluation cost of each block in the quadrant area of 32 × 32 pixels. At this time, the evaluation cost of the lower left block of 32 × 32 pixels is “25 = 10 + 5 + 5 + 5”, which is the sum of the evaluation costs of the lower left quadrant of 16 × 16 pixels. In other words, the final block selection unit 105 compares “100”, which is an evaluation cost of 64 × 64 pixels, with “85 = 25 + 25 + 20 + 15”, which is the total evaluation cost of each block in a quadrant of 32 × 32 pixels. To do. Since the total evaluation cost of the 32 × 32 pixel four-divided area is smaller, the final block selection unit 105 selects the block divided into 32 × 32 pixels. At this time, the final block selection unit 105 selects a block divided into 16 × 16 pixels in the lower left block of the block divided into 32 × 32 pixels. Based on the selected block, the final block selection unit 105 outputs the block structure shown in FIG.

  With the configuration of the first embodiment, the first candidate block determination unit 102 includes an LCU of 64 × 64 pixels, an LCU divided into 32 × 32 pixels, an LCU divided into 16 × 16 pixels, and 8 × 8 pixels. Based on the block size of each block of the LCU divided into two and the block size of the corresponding reference block, it is determined whether or not it corresponds to the first candidate block. The second candidate block determination unit 103 corresponds to the second candidate block based on the number of first candidate blocks occupying four blocks divided into four from a block having a large one-stage block size. It is determined whether or not. The block size evaluation value calculation unit 104 calculates the evaluation cost of the second candidate block, and the final block selection unit 105 selects a combination of blocks that minimizes the evaluation cost as a block constituting the block structure. As a result, it is possible to reduce the calculation amount of the block structure determination process while suppressing a decrease in encoding efficiency.

  When the above-described prior art 2 is applied to the example of FIG. 4, all the blocks of the LCU of 64 × 64 pixels, the LCU divided into 32 × 32 pixels, and the LCU divided into 16 × 16 pixels are candidates. The effect of reducing the amount of calculation is hardly obtained. On the other hand, in the configuration of the first embodiment of the present invention, since only the lower left block group is a candidate for the LCU divided into 16 × 16 pixels, the amount of calculation can be reduced.

  Similarly, when the above-described conventional technique 2 ′ is similarly applied to the example of FIG. 4, if the two blocks on the left and right of the LCU divided into 32 × 32 pixels have low correlation with the reference block, the conventional technique In 2 ′, the two blocks on the lower left and right are not candidates. For this reason, in the conventional technique 2 ', the coding efficiency is lowered by excluding these two blocks from the candidates. On the other hand, the configuration of the first embodiment of the present invention includes the second candidate block determination unit 103 that is not included in the block structure determination unit 1000 of the related art 2 'shown in FIG. Therefore, after each block is determined by the first candidate block determination unit 102, the second candidate block determination unit 103 further considers surrounding blocks (upper two blocks on the left and right shown in FIG. 6B). Thus, since the lower two blocks on the left and right are added to the candidates, a decrease in coding efficiency can be suppressed.

<Modification>
In the first embodiment, the condition shown in the condition (1) is a condition using a feature that the block size of the encoding target block is highly likely to be the same size as the block size of the reference block. is there. However, the configuration of the present invention is not limited to the embodiment, and the condition (1) is not limited to the above-described conditions, and the conditions shown in the prior art 1 and other conditions may be used.
Condition (2) is a condition that uses the feature that there is a block size correlation among four block groups divided into four blocks from a block size that is one step larger. However, the configuration of the present invention is not limited to the embodiment. The condition (2) is not limited to the above-described condition, and a plurality of blocks around the encoding target block may be used for the determination, and even in the LCU having the maximum block size of 64 × 64 pixels, You may make it use the surrounding block for a determination.

(Second embodiment)
FIG. 9 is a block diagram showing the configuration of the block structure determination unit 10a in the second embodiment of the present invention. The video encoding device 1 in the first embodiment includes a block structure determination unit 10. On the other hand, in the second embodiment, the block structure determining unit 10 is replaced with a block structure determining unit 10a. In FIG. 9, the same components as those in the first embodiment are denoted by the same reference numerals, and different configurations will be described below. The block structure determination unit 10 a includes an overlapping region processing unit 106 between the second candidate block determination unit 103 and the block size evaluation value calculation unit 104.

  The overlapping area processing unit 106 detects the overlapping area of the blocks determined to be the second candidate block by the second candidate block determination unit 103, and for the blocks that overlap more than a predetermined number of times. The second candidate block is excluded so as to satisfy the following condition (3). The excluded second candidate block is not included in the evaluation cost calculation target of the block size evaluation value calculation unit 104.

Condition (3): Excluded from the second candidate block so that the area occupied by the block satisfying the condition (1) is maximized for an area where the number of times of overlap is a predetermined number or more. However, when the proportions are the same, a small block size is preferentially excluded. In the present embodiment, the predetermined number of times is “2”.

  FIG. 10 is a flowchart showing the flow of the block structure determination process in the second embodiment. The process from step Sb1 to Sb3 is the same as the process from step Sa1 to Sa3 of the block structure determination process in the first embodiment. Through the processing up to step Sb3, the second candidate block determination unit 103 obtains the determination result shown in FIG. The overlap area processing unit 106 receives the output of the determination result from the second candidate block determination unit 103 and detects the number of overlaps in each block. FIG. 11A shows an example in which the overlap region processing unit 106 detects the number of overlaps in each block of a region divided into 32 × 32 pixels. For the upper left and right blocks and the lower right block, FIG. Since only the 64 × 64 pixel block area overlaps, “1” is detected as the number of overlaps. For the lower left block, in addition to the area of the block of 64 × 64 pixels, the area of the block divided into 16 × 16 pixels also overlaps, so “2” is detected as the number of overlaps.

  The lower left region that overlaps twice or more corresponds to a region in which the number of times of overlap of the condition (3) corresponds to a predetermined number of times or more, and thus becomes an exclusion target region as shown in FIG. The overlapping area processing unit 106 detects the ratio of the first candidate block in the lower left area. In the block of 64 × 64 pixels, the proportion of the first candidate block in the region is 100%. In the block of 16 × 16 pixels, the proportion of the first candidate block in the region is 75%. On the other hand, in the block of 32 × 32 pixels, the ratio of the first candidate block in the area is 0%. Therefore, the lower left block of 32 × 32 pixels having a low ratio is excluded. FIG. 12 is a diagram illustrating the determination result of each block after the processing by the overlapping area processing unit 106 is performed. In FIG. 12, “☆” indicates a block selected as the second candidate block, and “Δ "Indicates an excluded block.

  The block size evaluation value calculation unit 104 calculates the evaluation cost, that is, the RD cost, for each prediction mode, for each block that has become the second candidate block, excluding the block excluded from the second candidate block. 1) is calculated based on 1) (step Sb4). The numerical value in each block shown in FIG. 13A is the evaluation cost value calculated by the block size evaluation value calculation unit 104. The final block selection unit 105 selects a combination that minimizes the evaluation cost calculated by the block size evaluation value calculation unit 104 as a block constituting the block structure (step Sb5).

  The lower left area of the block divided into 32 × 32 pixels is excluded by the overlapping area processing unit 106. Therefore, the final block selection unit 105 applies the evaluation cost of the lower left four areas of the block divided into 16 × 16 pixels to the area, and calculates the total value of the evaluation costs as “85 = 25 + 20 + 15 + 10 + 5 + 5 + 5”. On the other hand, the evaluation cost of a block of 64 × 64 pixels is “100”. Therefore, the final block selection unit 105 is a combination as shown in FIG. 13B, that is, a block divided into 32 × 32 pixels in the upper left and lower right, and a block divided into 32 × 32 pixels in the lower left. A combination which is a block of four 16 × 16 pixels obtained by recursively dividing the generated block is selected. This combination is a combination of blocks constituting the block structure with the lowest evaluation cost.

In the configuration of the second embodiment, in addition to the configuration of the first embodiment, the overlap area processing unit 106 is provided between the second candidate block determination unit 103 and the block size evaluation value calculation unit 104. A configuration in which the overlapping area processing unit 106 excludes any overlapping block when the block determined as the second candidate block overlaps a predetermined number of times in the same area of the encoding target LCU. It was. As a result, it is possible to reduce the calculation amount of the evaluation cost by the block size evaluation value calculation unit 104 and increase the calculation speed while suppressing a decrease in encoding efficiency and reducing the calculation amount of the block structure determination process. .
In the example of the second embodiment described above, the block structure output by the final block selection unit 105 is the same as that of the first embodiment, but the calculation amount of the evaluation cost for one block of the LCU divided into 32 × 32 pixels. And the calculation speed is increased by reducing the amount of calculation.
In addition, the overlapping area processing unit 106 can leave a likely block as a candidate by making the ratio of the first candidate block the maximum.

In the second embodiment, the number of overlaps is used as a criterion for determination as the condition (3). However, the overlap area area, the ratio of the overlap area in the encoding target LCU, and the like are used instead of the number of overlaps. May be.
In the configuration of the second embodiment, the amount of calculation can be changed by a predetermined number of times used for determination of the number of times of duplication. In the example of FIG. 4, the number of times is set to “2”. In this case, the calculation amount reduction effect of 2/4 = 50% in the area ratio can be obtained. For example, consider a case where all blocks from 64 × 64 to 8 × 8 are the second candidate blocks. In this case, if the predetermined number of times is set to 2, blocks having an area of 2 LCUs are excluded. That is, since 4 LCUs are changed to 2 LCUs, the blocks for which the evaluation cost needs to be calculated are reduced by 50%. As a result, only 50% (for 2 LCUs) of the whole need to be calculated at the maximum. Further, if the overlap area area or the ratio of the overlap area in the encoding target LCU is used instead of the number of overlaps, the calculation amount can be changed with higher accuracy.

(Third embodiment)
Next, a third embodiment of the present invention will be described. FIG. 14 is a block diagram showing the configuration of the block structure determination unit 10b in the third embodiment of the present invention. The video encoding device 1 in the first embodiment includes a block structure determination unit 10. In contrast, in the third embodiment, the block structure determination unit 10 is replaced with the block structure determination unit 10b. In FIG. 14, the same components as those in the first embodiment are denoted by the same reference numerals, and different configurations will be described below. The block structure determination unit 10b includes a first candidate block determination unit 102b and a second candidate block determination unit 103b. The first candidate block determination unit 102b includes a 64 × 64 block determination unit 1021b, a 32 × 32 block determination unit 1022b, a 16 × 16 block determination unit 1023b, and an 8 × 8 block determination unit 1024b. The first candidate block determination unit 102b determines whether each block corresponds to the first candidate block for each prediction mode based on the LCU corresponding to each block size output by the block division unit 101. . In this embodiment, intra prediction mode and inter prediction mode are applied as prediction mode.

  First by each of the determination units of the first candidate block determination unit 102 (that is, 64 × 64 block determination unit 1021b, 32 × 32 block determination unit 1022b, 16 × 16 block determination unit 1023b, 8 × 8 block determination unit 1024b). The determination as to whether or not the candidate block is satisfied is performed as follows. In the case of the 64 × 64 block determination unit 1021b, one encoding target block, in the case of the 32 × 32 block determination unit 1022b, four encoding target blocks, and in the case of the 16 × 16 block determination unit 1023b, 16 In the case of the 8 encoding target blocks and the 8 × 8 block determination unit 1024b, 64 encoding target blocks are to be determined. For each of these encoding target blocks, whether or not the following condition (4) is satisfied is determined for each prediction mode.

  Condition (4): When the prediction mode is the intra prediction mode, when the block size of the block to be encoded is the same as the block size of the reference block or the same size as the block size of the reference block, The encoding target block is set as a first candidate block in the intra prediction mode. When the prediction mode is the inter prediction mode, when the block size of the encoding target block is the same as the block size of the reference block or the same size as the block size of the reference block, the encoding target block is Let it be the first candidate block in the inter prediction mode.

  Each determination unit of the first candidate block determination unit 102b outputs the determination result to the second candidate block determination unit 103b. The second candidate block determination unit 103b receives the determination result output from the first candidate block determination unit 102b and determines whether or not the following condition (5) is satisfied.

  Condition (5): When the encoding target block has a maximum block size (64 × 64 pixels in the present embodiment), a combination of a prediction mode and a block that satisfies the condition (4) satisfies the condition (5). Is determined as a second candidate block. If the block size is not the maximum block size, the second candidate block determination unit 103b pre-stores the block group belonging to the area obtained by dividing the block having a block size larger by one level into four in the internal storage area, based on the likelihood table shown in FIG. To calculate the total likelihood. When the calculated total likelihood value is equal to or greater than a predetermined likelihood threshold value, it is determined that the combination of the corresponding prediction mode and the block group satisfies the condition (5), and the blocks included in the block group All are the second candidate blocks in the prediction mode. At this time, in the plurality of prediction modes, when the calculated total likelihood value is equal to or greater than a predetermined likelihood threshold value, the prediction mode having the largest total value in the total likelihood values for each prediction mode. The block group corresponding to is set as the second candidate block. If the calculated total likelihood value is not equal to or greater than a predetermined likelihood threshold value, it is determined that the condition (5) is not satisfied, and the entire block group is not set as the second candidate block. However, the prediction mode with the smallest likelihood and the highest likelihood sum is given priority so that the region that has never become the second candidate block in the encoding target LCU at least once becomes the second candidate block. The second candidate block. In the present embodiment, a predetermined likelihood threshold value set in advance is “6”.

  Next, block structure determination processing in the third embodiment will be described with reference to FIGS. 16 to 20. FIG. 16 is a flowchart showing the flow of the block structure determination process. The block division unit 101 of the block structure determination unit 10 of the video encoding device 1 converts the LCU of the encoding target picture received from an external device or the like into block sizes of 32 × 32 pixels, 16 × 16 pixels, and 8 × 8 pixels. To divide. The block dividing unit 101 includes an LCU having a maximum block size of 64 × 64 pixels, an LCU divided into 32 × 32 pixels, an LCU divided into 16 × 16 pixels, and an LCU divided into 8 × 8 pixels. Is output to the corresponding determination unit of the first candidate block determination unit 102b (step Sc1).

  The first candidate block determination unit 102b determines whether or not the encoding target block is a first candidate block that satisfies the condition (4), and performs a process of selecting the first candidate block (step Sc2). For example, the processing of step Sc2 is performed by each determination unit in the order of 64 × 64 block determination unit 1021b, 32 × 32 block determination unit 1022b, 16 × 16 block determination unit 1023b, and 8 × 8 block determination unit 1024b. This is performed for the LCU having the corresponding block size (loop Lc1). Each determination unit of the first candidate block determination unit 102 repeatedly performs the process of step Sc2 on each of the blocks included in the encoding target LCU (loop Lc2). For example, in the case of the 8 × 8 block determination unit 1024b, step Sa2 is repeated 64 times for each of the 64 blocks. Each determination unit of the first candidate block determination unit 102b performs the process of step Sc2 for each prediction mode, that is, in the present embodiment, for the intra prediction mode and the inter prediction mode (loop Lc3). When the process of step Sc2 is performed for each prediction mode by each determination unit of the first candidate block determination unit 102b, the loop Lc3 ends (loop Lc4). Further, when the determining unit of the first candidate block determining unit 102b performs the process of step Sc2 on each of the blocks included in the encoding target LCU, the loop Lc2 ends (loop Lc5). Further, when the determination unit of the first candidate block determination unit 102b performs the process of step Sc2 on the LCUs having the corresponding block size, the loop Lc1 ends (loop Lc6).

  FIG. 17 is a diagram illustrating a determination result by the first candidate block determination unit 102b. In FIG. 17, “◯” indicates that the condition (4) is satisfied, and “X” indicates that the condition (4) is not satisfied. Since the block of 64 × 64 pixels has the maximum block size, it satisfies the condition (4) regardless of the prediction mode, and therefore becomes the first candidate block as shown in FIG. . For the LCU divided into 32 × 32 pixels, the intra prediction mode has the same result as in the first embodiment, the upper left and right blocks are the first candidate blocks, and the lower left and right blocks are the first candidates. It is not a block. On the other hand, in the inter prediction mode, since the size of the reference block overlapping the upper right block is 64 × 64 pixels, the condition (4) is not satisfied. The upper left block has the same block size as the reference block, and the lower left and right blocks satisfy the condition (4) because the reference block is 16 × 16 pixels, and thus are the first candidate blocks. Therefore, the determination result of the LCU divided into 32 × 32 pixels by the first candidate block determination unit 102b is as shown in FIG.

  For the LCU divided into 16 × 16 pixels, the same block as the first embodiment is the first candidate block in the intra prediction mode. On the other hand, in the inter prediction mode, since the center coordinates of the four blocks in the fourth row of the LCU are all in a 16 × 16 pixel block or a 32 × 32 pixel block, the condition (4 ) To satisfy the first candidate block. Therefore, the determination result of the LCU divided into 16 × 16 pixels by the first candidate block determination unit 102b is as shown in FIG. It is determined that all the 8 × 8 pixels do not satisfy the condition (4).

  Next, the second candidate block determination unit 103b receives the determination result output from the first candidate block determination unit 102b, and determines the second candidate block based on the condition (5) (step Sc3). The second candidate block determination unit 103b has a total value of the likelihood of the four regions of the LCU divided into 4 × 32 × 32 pixels and a 16 × 16 pixel each of which is further divided into 4 blocks of 32 × 32 pixels. The total likelihood value of the four regions is calculated for each prediction mode using the likelihood table shown in FIG. FIG. 18 is a calculation result of the total likelihood value by the second candidate block determination unit 103b. For a 32 × 32 pixel LCU, the total likelihood value is “11”. The total likelihood value of the four left upper regions of 16 × 16 pixels is “6”, the total likelihood value of the four lower left regions of 16 × 16 pixels is “13”, and 16 × The total likelihood value of the four pixels at the lower right of 16 pixels is “10”.

  The second candidate block determination unit 103b determines whether or not the total likelihood value corresponds to the second candidate block based on whether or not the total likelihood value is a predetermined likelihood threshold, that is, “7” or more. To do. As shown in FIG. 19B, since the LCU divided into 32 × 32 pixels has a total likelihood value of “11”, the second candidate block determination unit 103b satisfies the condition (5). Is determined. As illustrated in FIG. 19B, the second candidate block determination unit 103b has a total likelihood value of “2” for the intra prediction mode, and a total likelihood value of “9” for the inter prediction mode. Is. Since the inter prediction mode has the highest likelihood total value, the second candidate block determination unit 103b sets the upper right block that is “x” in the inter prediction mode as the second candidate block.

  As shown in FIG. 19 (c), for the four blocks at the upper left of the LCU divided into 16 × 16 pixels, the total likelihood value is “6”, so the second candidate block determination unit 103b Not all four blocks are second candidate blocks.

  For the four blocks at the lower left, the total likelihood value in the intra prediction mode is “9” and the total likelihood value in the inter prediction mode is “4”, so the total likelihood value as a whole Is “13”, which is equal to or greater than the likelihood threshold. Since the total likelihood value is larger in the intra prediction mode, the second candidate block determination unit 103b selects the upper right block that is “x” for the block in the intra prediction mode as the second candidate. Let it be a block.

  For the four blocks on the lower right, the total likelihood value in the intra prediction mode is “6”, the total likelihood value in the inter prediction mode is “4”, and the total likelihood value as a whole Is “10”, which is equal to or greater than the likelihood threshold. Since the total likelihood value is larger in the intra prediction mode, the second candidate block determination unit 103b determines the second block on the left and right in the intra prediction mode as the second block. Candidate block. Finally, the second candidate block determination unit 103b selects the second candidate block as shown in FIG. 19C.

  The block size evaluation value calculation unit 104 calculates an evaluation cost, that is, an RD cost for each prediction mode for each block that has become the second candidate block based on the above equation (1) (step Sc4). A numerical value in each block shown in FIG. 20A is an evaluation cost value calculated by the block size evaluation value calculation unit 104. The final block selection unit 105 selects a combination that minimizes the evaluation cost calculated by the block size evaluation value calculation unit 104 as a block constituting the block structure (step Sc5).

  As shown in FIG. 20A, first, the final block selection unit 105 evaluates “30” that is the evaluation cost of the lower left block of 32 × 32 pixels and the lower left quadrant area of 16 × 16 pixels. The total cost is compared with “25 = 10 + 5 + 5 + 5”. Since the total evaluation cost of the lower left quadrant area of 16 × 16 pixels is smaller, the final block selection unit 105 selects a block divided into 16 × 16 pixels. Further, the final block selection unit 105 sets “15” which is the evaluation cost of the lower right block of 32 × 32 pixels and “14 == the total evaluation cost of the lower right quadrant of 16 × 16 pixels”. 5 + 3 + 3 + 3 ”. Since the total evaluation cost of the lower left quadrant area of 16 × 16 pixels is smaller, the final block selection unit 105 selects a block divided into 16 × 16 pixels.

  Next, the final block selection unit 105 compares “100”, which is the evaluation cost of 64 × 64 pixels, with the total evaluation cost of each block in the quadrant area of 32 × 32 pixels. At this time, the evaluation cost of the lower left block of 32 × 32 pixels is “25 = 10 + 5 + 5 + 5”, which is the sum of the evaluation costs of the lower left quadrant of 16 × 16 pixels. The evaluation cost of the lower right block of 32 × 32 pixels is “14 = 5 + 3 + 3 + 3”, which is the total evaluation cost of the lower right quadrant of 16 × 16 pixels. That is, the final block selection unit 105 compares “100”, which is an evaluation cost of 64 × 64 pixels, with “84 = 25 + 20 + 25 + 14”, which is the total evaluation cost of each block in a quadrant of 32 × 32 pixels. To do.

  Since the total evaluation cost of the 32 × 32 pixel four-divided area is smaller, the final block selection unit 105 selects the block divided into 32 × 32 pixels. At this time, the final block selection unit 105 selects a block divided into 16 × 16 pixels in the lower left block of the block divided into 32 × 32 pixels. The final block selection unit 105 selects a block divided into 16 × 16 pixels in the lower right block of the block divided into 32 × 32 pixels. Based on the selected block, the final block selection unit 105 outputs the block structure shown in FIG.

  With the configuration of the third embodiment, the first candidate block determination unit 102b includes an LCU of 64 × 64 pixels, an LCU divided into 32 × 32 pixels, an LCU divided into 16 × 16 pixels, and 8 × 8 pixels. Based on the block size of each block of the LCU divided into two and the block size of the corresponding reference block, it is determined whether the block corresponds to the first candidate block according to the type of prediction mode. The second candidate block determination unit 103 has a likelihood determined in advance according to the prediction mode and the block size, and a ratio of the first candidate block to four blocks divided into four blocks from a block having a large one-stage block size. Based on the above, the total likelihood value of the four blocks is calculated. When the total likelihood value is equal to or greater than a predetermined likelihood threshold, the block group is set as the second candidate block. As a result, it is possible to reduce the calculation amount of the block structure determination process while suppressing a decrease in encoding efficiency.

  In the configuration of the first embodiment, it is determined whether or not each block is the first candidate block and the second candidate block. Therefore, when a block that is not the first candidate block is added as the second candidate block by the determination process of the second candidate block determination unit 103, it is necessary to perform calculations for all applied prediction modes. On the other hand, in the configuration of the third embodiment, since the determination is performed in consideration of the prediction mode, the determination accuracy is improved, and the likelihood value is further increased when the second candidate block is added. Only the block corresponding to the large prediction mode is set as the second candidate block. Therefore, it is only necessary to calculate the prediction mode for the added block, and the second candidate block can be added while suppressing an increase in the calculation amount. For example, in the example of FIG. 4, in the third embodiment, the lower right block group of the LCU divided into 16 × 16 pixels, which is not the second candidate block in the first embodiment, is newly added as the second candidate. Although it is added as a block, two of the four blocks calculate the evaluation cost only in the inter prediction mode. It is possible to output a high block structure.

In the third embodiment, the condition (4) is a condition using a feature that the likelihood differs depending on the prediction mode in the candidate block size. However, the configuration of the present invention is not limited to this embodiment, and conditions other than those described above may be applied.
Moreover, conditions other than those described above may also be applied to the condition (5). For example, the likelihood table and the likelihood threshold shown in FIG. 15 may be adaptively changed according to the encoding result. . In consideration of the positional relationship of blocks, for example, the prediction mode of a block that is newly set as a second candidate block may be set to a prediction mode with a large number of prediction mode candidates of adjacent blocks. Further, when the same prediction mode is a candidate in an adjacent block, the likelihood value may be set large in the likelihood table shown in FIG.

  In the first to third embodiments, the configuration conforming to HEVC is targeted, but the configuration is not limited to the configuration conforming to HEVC.

  In the first to third embodiments, the LCU size is 64 × 64 pixels, and the CU size is 4 × 64 pixels, 32 × 32 pixels, 16 × 16 pixels, and 8 × 8 pixels. A block constituting the block structure is selected from the pattern. However, the configuration of the present invention is not limited to this embodiment, and can also be applied to an embodiment in which blocks constituting a block structure are selected from two or more patterns.

  In the first to third embodiments, one motion vector is shown as a motion vector indicating the relationship between the encoding target LCU described in FIG. 3 and the reference block. Instead, the block located in the same position as the encoding target LCU may be used as a reference block, and the motion vector of each block of the encoding target LCU is used instead of the motion vector for each encoding target LCU. May be.

  In the first to third embodiments, the RD cost calculated by the block size evaluation value calculation unit 104 is an example of the evaluation cost, and any index can be used for evaluating the block. For example, a statistical value such as only an error, only a code amount, or a variance value of luminance values may be applied as the evaluation cost.

  Moreover, in said 1st embodiment to 3rd embodiment, although intra prediction mode and inter prediction mode are applied as prediction mode, embodiment of this invention is limited to the said embodiment. Absent. In addition to the intra prediction mode and the inter prediction mode, the Merge prediction mode and the Skip prediction mode may be applied, or any two or more of these prediction modes may be selected and applied. .

For example, when the Skip prediction mode and the Merge prediction mode are applied, if the reference block is the Skip prediction mode, the first block size is the same as the block size of the reference block or the block size that is one step larger than the block size of the reference block. Let it be one candidate block. When the reference block is other than the Skip prediction mode, the first candidate block is determined if it is the same as the block size of the reference block or the block size that is one step smaller than the block size of the reference block.
As a method of selecting the second candidate block, for example, in the four blocks divided into four, if the ratio of the first candidate block is 50% or more, the remaining blocks that are not the first candidate block are included. There is a technique in which all are used as second candidate blocks. As another method, there is also a method of selecting the second candidate block by the score. For example, when the block size of the reference block is 32 × 32 pixels and the Skip mode is used, there is a high probability that the encoding target block is also 32 × 32 pixels, and when the prediction mode of the reference block is the inter prediction mode, The probability that the encoding target block is 32 × 32 pixels is normal, 1 point, and when the prediction mode of the reference block is the intra prediction mode, the probability that the encoding target block is 32 × 32 pixels is low, 0 points. Determine the score as follows. There is also a method of selecting all four blocks as second candidate blocks when the number of points in the four blocks divided into four exceeds four points.

  Moreover, in said 1st embodiment to 3rd embodiment, the 2nd candidate block determination part 103,103b has determined whether the divided | segmented block group is made into a 2nd candidate block collectively, By determining a plurality of blocks in this way, a device that specializes in parallel processing such as a GPU (Graphics Processing Unit) can achieve the effect of speeding up by processing a large amount of data collectively. There are advantages.

  In the first to third embodiments, the loop processing of La1, La2, Lb1, Lb2, Lc1, Lc2, and Lc3 in the flowcharts of FIGS. 5, 10, and 16 is performed in order as described above. It may be performed, or processing may be performed in parallel.

Moreover, in said 1st embodiment to 3rd embodiment, the determination made into more than the predetermined number of conditions (2) in 1st embodiment, for example, the determination made more than the predetermined number in the determination of the frequency | count of duplication in 2nd embodiment In the third embodiment, the process of determining to be greater than or equal to the likelihood threshold is an example, and may be a determination to exceed a certain value and to be less than or equal to a certain value, instead of determining to be greater than a certain value and less than a certain value. .
Moreover, in said 1st embodiment to 3rd embodiment, since all 8 * 8 pixels do not satisfy | fill conditions (1), determination that all 8 * 8 pixels do not correspond to a 1st candidate block is made. Although an example has been shown, in the following cases, a case where a part of 8 × 8 pixels becomes the first candidate block is also assumed. In the case of 8 × 8, the center coordinates are present in a 16 × 16 pixel block of the reference block. In this case, since the size is one step smaller than the size of the reference block, a part of 8 × 8 pixels becomes the first candidate block.

  You may make it implement | achieve the block structure determination part 10, 10a, 10b in embodiment mentioned above with a computer. In that case, a program for realizing this function may be recorded on a computer-readable recording medium, and the program recorded on this recording medium may be read into a computer system and executed. Here, the “computer system” includes an OS and hardware such as 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” dynamically holds a program for a short time like 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 serving as a server or a client in that case may be included and a program held for a certain period of time. Further, the program may be a program for realizing a part of the above-described functions, and may be a program capable of realizing the functions described above in combination with a program already recorded in a computer system. You may implement | achieve using programmable logic devices, such as FPGA (Field Programmable Gate Array).

  The embodiment of the present invention has been described in detail with reference to the drawings. However, the specific configuration is not limited to this embodiment, and includes designs and the like that do not depart from the gist of the present invention.

DESCRIPTION OF SYMBOLS 1 ... Video coding apparatus, 10 ... Block structure determination part, 11 ... Subtractor, 12 ... Orthogonal transformation / quantization part, 13 ... Variable length coding part, 14 ... Dequantization / inverse orthogonal transformation part, 15 ... Addition 16 ... Intra prediction unit, 17 ... Intra / inter switch, 18 ... Loop filter unit, 19 ... Decoded picture memory unit, 20 ... Inter prediction unit, 101 ... Block division unit, 102 ... First candidate block determination unit, DESCRIPTION OF SYMBOLS 103 ... 2nd candidate block determination part, 104 ... Block size evaluation value calculation part, 105 ... Final block selection part, 1021 ... 64x64 block determination part, 1022 ... 32x32 block determination part, 1023 ... 16x16 block determination Part, 1024 ... 8 × 8 block determination part

Claims (8)

Dividing an image into a plurality of regions, and determining a block structure configured by one block size generated by recursively dividing the region into blocks or a combination of blocks having a plurality of block sizes A block structure determination unit;
An encoding unit that performs encoding based on the determined block structure;
With
The block structure determining unit
A block dividing unit that divides the area of the image into blocks of a predetermined size for each block size;
If the size of the divided block is equal to the size of the block to which the center of the area referred to by the block belongs, or if the size of the block to which the center belongs is one step larger, each of the blocks is the first candidate a first candidate block determination unit determines that the block,
When the number of blocks determined by dividing a block having a size one step larger than the predetermined size into the predetermined size is equal to or greater than a predetermined value, the block group all block and the second candidate block determination unit determines to a second candidate block belonging,
A block size evaluation value calculation unit for calculating an evaluation value for the block determined as the second candidate block;
A final block selection unit that selects a combination of the blocks that divide the image area based on the evaluation value;
A video encoding device comprising: The block dividing unit divides the image into a plurality of predetermined sizes,
The first candidate block determination unit and the second candidate block determination unit perform determination for each of the plurality of sizes,
When it is determined by the second candidate block determination unit that the predetermined region of the image overlaps with the second candidate block, any block of the predetermined region of the image is set as the second candidate block. Exclude from candidate block
The video encoding apparatus according to claim 1, further comprising an overlapping area processing unit . The first candidate block determination unit, for each prediction mode applied to the image, based on the divided block and a reference block that divides a region in a reference picture corresponding to the region of the image, Determine whether each of the blocks is a first candidate block;
The second candidate block determination unit is included in the block group for each prediction mode with respect to a block group obtained by dividing a block having a size one step larger than the predetermined size into the predetermined size. Based on the ratio of the first candidate block and the likelihood determined in advance according to the prediction mode and the block size, it is determined whether or not all the blocks in the block group are the second candidate blocks. The video encoding device according to claim 1 or 2 . The first candidate block determination unit is configured such that, for any one of the prediction modes, the block size of each of the divided blocks is the same as the block size of the reference block, or one stage than the block size of the reference block If the block size is the same as the small block size, the block size of each of the divided blocks is the same as the block size of the reference block or the reference block for the prediction mode other than the prediction mode as the first candidate block The video encoding device according to claim 3 , wherein when the block size is the same as a block size that is one step larger than the block size, the first candidate block is determined. Different likelihoods are predetermined for each combination of block size and the prediction mode,
The second candidate block determination unit calculates the likelihood of the block group for each prediction mode based on the block size of the block group and the prediction mode, and based on the calculated total likelihood value The video encoding device according to claim 3 , wherein it is determined whether or not all blocks of the block group are second candidate blocks. When the second candidate block determining unit determines that all the blocks in the block group are the second candidate blocks, the block that is not the first candidate block in the block group is added as the second candidate block. In this case, the block is added as the second candidate block corresponding to the prediction mode having a large total likelihood value among the total likelihood values obtained in each of a plurality of prediction modes. 5. The video encoding device according to 5. A block structure configured by one block size generated by dividing an image region into blocks and recursively dividing the region into blocks or a combination of blocks having a plurality of block sizes is determined. A block structure determination step;
An encoding step for performing encoding based on the determined block structure;
Have
In the block structure determination step,
A block dividing step of dividing the image area into blocks of a predetermined size for each block size;
If the size of the divided block is equal to the size of the block to which the center of the area referred to by the block belongs, or if the size of the block to which the center belongs is one step larger, each of the blocks is the first candidate a first candidate block determination step of determining that the block,
When the number of blocks determined by dividing a block having a size one step larger than the predetermined size into the predetermined size is equal to or greater than a predetermined value, the block group a second candidate block determination step determines to all blocks belonging to the second candidate block,
A block size evaluation value calculating step for calculating an evaluation value for the block determined to be the second candidate block;
A final block selection step of selecting a combination of the blocks that divide the region of the image based on the evaluation value;
A block structure determination method. A block structure configured by one block size generated by dividing an image region into blocks and recursively dividing the region into blocks or a combination of blocks having a plurality of block sizes is determined. A block structure determination step;
An encoding step for performing encoding based on the determined block structure;
To the computer,
In the block structure determination step,
A block dividing step of dividing the image area into blocks of a predetermined size for each block size;
If the size of the divided block is equal to the size of the block to which the center of the area referred to by the block belongs, or if the size of the block to which the center belongs is one step larger, each of the blocks is the first candidate a first candidate block determination step of determining that the block,
When the number of blocks determined by dividing a block having a size one step larger than the predetermined size into the predetermined size is equal to or greater than a predetermined value, the block group a second candidate block determination step determines to all blocks belonging to the second candidate block,
A block size evaluation value calculating step for calculating an evaluation value for the block determined to be the second candidate block;
A final block selection step of selecting a combination of the blocks that divide the region of the image based on the evaluation value;
A computer program for running.

Images