JP2015139117A - Information processing apparatus, selection method of coding unit, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce an arithmetic load relating to prediction coding.SOLUTION: There is provided an information processing apparatus 10 which includes: a storage part 11 for storing an object block BLo and an adjacent block BLn adjacent to the object block BLo; and an arithmetic part 12 which evaluates similarity between a first area A1, in the object block BLo, in contact with a boundary with the adjacent block BLn and an adjacent area An, in the adjacent block BLn, in contact with the boundary, evaluates similarity between the first area A1 and a second area A2, in the object block BLo, in contact with the first area A1, and defines the first area A1 and an area A5 excluding the first area A1 from the inside of the object block BLo as candidates of a unit of prediction coding in the case where the similarity between the first area A1 and the adjacent area An is higher than the similarity between the first area A1 and the second area A2.

Description

  The present invention relates to an information processing apparatus, a coding unit selection method, and a program.

  In the fields of digital broadcasting and video recording, AVC (Advanced Video Coding) / H. A video compression technique (hereinafter referred to as AVC method) defined by a standard called H.264 is widely used. In recent years, HEVC (High Efficiency Video Coding) / H. Attention has been focused on a moving picture compression technology called H.265 (hereinafter referred to as HEVC method). The HEVC method aims to realize a compression rate twice that of the AVC method.

  In order to realize the high compression rate as described above, in the HEVC system, a device has been added to give flexibility in how to determine a block as a unit of encoding. In the AVC method, a picture is divided in units of 16 × 16 pixel macroblocks, and the macroblocks are used as encoding units. In the HEVC scheme, a picture is divided in units of fixed-size blocks called CTU (Coding Tree Unit), and the CTU is further divided into units called CU (Coding Unit). This CU is a unit of encoding.

  The division of the CTU is performed recursively according to the image characteristics. That is, the CTU is divided into four blocks, and the process of dividing the block obtained by the division into four blocks is repeatedly performed to determine a CU having a size suitable for the feature of the image. The size of the CU is 64 × 64 pixels (CU depth 0), 32 × 32 pixels (CU depth 1), 16 × 16 pixels (CU depth 2), or 8 × 8 pixels (CU depth 3). is there. Such a division mechanism is sometimes referred to as quad tree block division.

  By adopting quadtree block division, for example, it is possible to encode an edge portion such as the contour of an object and its vicinity in small CU units, and to encode a flat portion such as the sky in large CU units. . As a result, it is possible to achieve high image quality while suppressing the code amount. In the HEVC scheme, inter-frame prediction encoding (hereinafter, inter prediction) is performed in units of blocks called PUs (Prediction Units) set for each CU. Each CU is set with at least one PU.

  PU is the same area as CU (no division), CU is divided into 2 parts in vertical or horizontal direction, CU is divided in 1 to 3 or 3 to 1 ratio in the vertical or horizontal direction (hereinafter asymmetric division) ) Is set to one of the areas. That is, PUs obtained by seven types of division methods can be used for inter prediction. In addition, since there are four types of CU sizes as described above, in the HEVC scheme, a total of 28 types of PUs can be used for inter prediction.

  Hereinafter, the PU set in the same area as the CU is denoted as PU [2N × 2N]. In addition, a PU in a region obtained by dividing the CU into two in the vertical direction is denoted as PU [N × 2N]. In addition, a PU in a region obtained by dividing the CU into two in the horizontal direction is denoted as PU [2N × N]. Also, in the region in which the CU is asymmetrically divided in the vertical direction, the left PU is expressed as PU [nL × 2N], and the right PU is expressed as PU [nR × 2N]. In addition, in the region in which the CU is asymmetrically divided in the horizontal direction, the upper PU is expressed as PU [2N × nU], and the lower PU is expressed as PU [2N × nD].

As described above, in the HEVC scheme, since unit blocks (PUs) of various sizes can be used for inter prediction, it is possible to compress moving images more efficiently than in the AVC scheme.
By the way, with respect to inter prediction, a method of calculating a spatial similarity between a current block and a plurality of adjacent partitions and estimating a motion vector of at least one adjacent partition selected based on the spatial similarity as a motion vector of the current block. Has been proposed. In addition, at least two prediction modes are selected based on the cost value generated from the difference information between the prediction signal and the video signal, and the optimum prediction mode is selected based on the frequency information of the difference information regarding the selected prediction mode. A method has been proposed.

Special table 2010-515399 gazette JP 2006-180195 A

  As described above, the accuracy of inter prediction can be increased by using PUs of various sizes. On the other hand, since there are 28 types of PUs, the load on the process of selecting a suitable PU type is heavy. Therefore, according to one aspect, an object of the present invention is to provide an information processing apparatus, a coding unit selection method, and a program that can reduce a calculation load related to predictive coding.

  According to one aspect of the present disclosure, a storage unit that stores a target block and an adjacent block adjacent to the target block, a first area that is in contact with the boundary of the adjacent block in the target block, and an intermediate block To evaluate the similarity between the adjacent area in contact with the boundary and evaluate the similarity between the first area and the second area in contact with the first area in the target block, and the similarity between the first area and the adjacent area Is higher than the similarity between the first region and the second region, the first region and a calculation unit that uses the region excluding the first region from the target block as a candidate for a predictive coding unit, An information processing apparatus is provided.

  According to another aspect of the present disclosure, a computer that can read out information on a target block and an adjacent block from a storage unit that stores the target block and an adjacent block adjacent to the target block. The first area that is in contact with the boundary between adjacent blocks and the adjacent area that is in contact with the boundary among adjacent blocks is evaluated, and the first area and the second area that is in contact with the first area in the target block When the similarity between the first area and the adjacent area is higher than the similarity between the first area and the second area, the first area and the first area among the target blocks are evaluated. There is provided a method for selecting a coding unit in which a region excluding is used as a candidate for a predictive coding unit.

  According to another aspect of the present disclosure, a computer that can read out information on a target block and an adjacent block from a storage unit that stores the target block and an adjacent block adjacent to the target block. The first area that is in contact with the boundary between adjacent blocks and the adjacent area that is in contact with the boundary among adjacent blocks is evaluated, and the first area and the second area that is in contact with the first area in the target block When the similarity between the first area and the adjacent area is higher than the similarity between the first area and the second area, the first area and the first area among the target blocks are evaluated. There is provided a program that executes a process in which a region excluding the region is a candidate for a predictive coding unit.

  According to the present invention, it is possible to reduce the calculation load for predictive coding.

It is the figure which showed an example of the information processing apparatus which concerns on 1st Embodiment. It is a figure for demonstrating the kind of CU. It is a figure for demonstrating the kind of PU. It is a figure for demonstrating the motion detection based on PU. It is a 1st flowchart for demonstrating the division | segmentation process of CU. It is a 2nd flowchart for demonstrating the division | segmentation process of CU. It is the figure which showed an example of the hardware which can implement | achieve the function which the information processing apparatus which concerns on 2nd Embodiment has. It is the block diagram which showed an example of the function which the information processing apparatus which concerns on 2nd Embodiment has. It is a figure for demonstrating the division | segmentation method (vertical direction division | segmentation) of PU which concerns on 2nd Embodiment. It is a figure for demonstrating the division | segmentation method (horizontal direction division | segmentation) of PU which concerns on 2nd Embodiment. It is a figure for demonstrating the evaluation of the similarity which concerns on 2nd Embodiment. It is a 1st flowchart for demonstrating the division | segmentation process of CU which the information processing apparatus which concerns on 2nd Embodiment performs. It is a 2nd flowchart for demonstrating the division | segmentation process of CU which the information processing apparatus which concerns on 2nd Embodiment performs. It is a 3rd flowchart for demonstrating the division | segmentation process of CU which the information processing apparatus which concerns on 2nd Embodiment performs. It is the block diagram which showed an example of the function which the information processing apparatus which concerns on 3rd Embodiment has. It is a figure for demonstrating the division | segmentation method (vertical direction division | segmentation) and similarity evaluation which concern on 3rd Embodiment. It is a figure for demonstrating the division | segmentation method (horizontal direction division | segmentation) and similarity evaluation which concern on 3rd Embodiment.

  Embodiments of the present invention will be described below with reference to the accompanying drawings. In addition, about the element which has the substantially same function in this specification and drawing, duplication description may be abbreviate | omitted by attaching | subjecting the same code | symbol.

<1. First Embodiment>
The first embodiment will be described with reference to FIG. FIG. 1 is a diagram illustrating an example of an information processing apparatus according to the first embodiment. Note that the information processing apparatus 10 illustrated in FIG. 1 is an example of an information processing apparatus according to the first embodiment.

As illustrated in FIG. 1, the information processing apparatus 10 includes a storage unit 11 and a calculation unit 12.
The storage unit 11 is a volatile storage device such as a RAM (Random Access Memory) or a non-volatile storage device such as an HDD (Hard Disk Drive) or a flash memory. The arithmetic unit 12 is a processor such as a CPU (Central Processing Unit) or a DSP (Digital Signal Processor). However, the arithmetic unit 12 may be an electronic circuit such as an application specific integrated circuit (ASIC) or a field programmable gate array (FPGA). For example, the calculation unit 12 executes a program stored in the storage unit 11 or another memory.

  The storage unit 11 stores the target block BLo and the adjacent block BLn adjacent to the target block BLo. The target block BLo is, for example, an encoding target CU. The adjacent block BLn is, for example, a CU adjacent to the target block BLo. In the example of FIG. 1, for convenience of explanation, only one adjacent block BLn positioned above the target block BLo is shown. Of course, the technique according to the first embodiment can be similarly applied to blocks adjacent to the lower side, the left side, and the right side of the target block BLo.

  The computing unit 12 evaluates the similarity between the first area A1 in contact with the boundary with the adjacent block BLn in the target block BLo and the adjacent area An in contact with the boundary in the adjacent block BLn. In the example of FIG. 1, the size of the target block BLo is N × N pixels. The size of the first area A1 and the adjacent area An is N × (N / 4) pixels. For example, the calculation unit 12 calculates the sum of absolute differences between the luminance of the first region A1 and the luminance of the adjacent region An for each pixel. And the calculating part 12 evaluates that similarity is so high that a difference absolute value sum is small.

  In addition, the calculation unit 12 evaluates the similarity between the first area A1 and the second area A2 in contact with the first area A1 in the target block BLo. In the example of FIG. 1, the size of the second area A2 is N × (N / 4) pixels. For example, the calculation unit 12 calculates the sum of absolute differences between the luminance of the first region A1 and the luminance of the second region A2 for each pixel. And the calculating part 12 evaluates that similarity is so high that a difference absolute value sum is small. Note that although a method of using the sum of absolute differences of luminance for similarity evaluation is illustrated, a method of using a color difference or a method of using flatness of an area is also applicable.

  When the similarity between the first area A1 and the adjacent area An is higher than the similarity between the first area A1 and the second area A2, the calculation unit 12 selects the first area A1 and the target block BLo from the first area A1 and the target block BLo. A region A5 excluding the first region A1 is set as a candidate for a predictive coding unit. In addition, when the similarity between the first region A1 and the adjacent region An is lower than the similarity between the first region A1 and the second region A2, the calculation unit 12 determines whether the first region A1 and the second region A2 A third region A3 obtained by combining the two is set as a candidate for a predictive coding unit. Further, the calculation unit 12 sets a region A4 obtained by removing the third region A3 from the target block BLo as a candidate for a predictive coding unit.

  For example, the evaluation height for the similarity between the first region A1 and the second region A2 is expressed as EV12, and the evaluation height for the similarity between the first region A1 and the adjacent region An is expressed as EV1n. The calculation unit 12 determines whether or not the evaluation EV12 is higher than the evaluation EV1n. When the evaluation EV12 is lower than the evaluation EV1n, the calculation unit 12 sets the first region A1 and the region A5 as prediction encoding unit candidates. On the other hand, when the evaluation EV12 is higher than the evaluation EV1n, the calculation unit 12 sets the third region A3 and the region A4 as candidates for prediction encoding units. Note that the unit of predictive coding is, for example, PU.

  For example, when the above mechanism is applied when there are a method of selecting a prediction encoding unit, a method of dividing the target block BLo into two parts vertically and a method of dividing the target block BLo up and down at a ratio of 1: 3, Narrow down to one method based on similarity assessment. For example, when the unit of predictive coding is a PU, the types of PUs to be evaluated based on motion detection are narrowed down. As a result, the burden of processing for selecting a suitable PU is reduced, and the calculation load for predictive coding is reduced. It should be noted that the effect is further enhanced by executing the same processing for adjacent blocks located below, on the right side, and on the left side of the target block BLo.

The first embodiment has been described above.
<2. Second Embodiment>
Next, a second embodiment will be described. The second embodiment relates to a PU determination method (CU division method) in the HEVC scheme.

[2-1. CU division method]
A method of dividing a CU will be described with reference to FIGS.
(CU type (CU depth))
In the HEVC method, a picture having a luminance component and a color difference component is divided in CTU units of a fixed size, and encoding and decoding are performed in CTU units. The CTU includes a luminance component CTB (Coding Tree Block) and a color difference component CTB. CTB is a sub-block that is a processing unit such as DCT (Discrete Cosine Transform) conversion and quantization. There are three types of CTB sizes: 16 × 16 pixels, 32 × 32 pixels, and 64 × 64 pixels. The CTU is hierarchically divided in a quadtree format as shown in FIG. FIG. 2 is a diagram for explaining the types of CUs.

  The depth of the hierarchy is expressed by “CU depth”. The highest hierarchy is expressed as “CU depth 0”. Hierarchies lower than “CU depth 0” are expressed in order as “CU depth 1”, “CU depth 2”, and “CU depth 3”. If the CTU size is 64 × 64, the CU depth 0 CU has a size of 64 × 64. The CU depth 1 CU has a size of 32 × 32. The CU depth 2 CU has a size of 16 × 16. The CU depth 3 CU has a minimum size of 8 × 8. The minimum size of the CU is fixed throughout the sequence.

(PU type)
PU is set to the area | region which divided | segmented CU. There are seven types of division methods (including the case of “no division”) as shown in FIG. FIG. 3 is a diagram for explaining the types of PUs. (1) in FIG. 3 is a method in which the entire CU is a PU without dividing the CU. When the size of the CU is 2N × 2N pixels, the size of the PU is also 2N × 2N pixels. This type is denoted as PU [2N × 2N]. (2) in FIG. 3 is a method in which two regions obtained by dividing a CU into two in the horizontal direction are used as PUs. The size of the PU is N × 2N pixels. This type is expressed as PU [N × 2N].

  (3) in FIG. 3 is a method in which two regions obtained by dividing a CU into two in the vertical direction are used as PUs. The size of the PU is 2N × N pixels. This type is denoted as PU [2N × N]. (4) in FIG. 3 is a method in which two regions obtained by dividing a CU into two in the horizontal direction from the right side at a ratio of 1: 3 are used as PUs. The size of the PU located on the right side is (N / 2) × 2N pixels. On the other hand, the size of the PU located on the left side is (3N / 2) × 2N pixels. This type is denoted as PU [nR × 2N].

  (5) in FIG. 3 is a method in which two regions obtained by dividing a CU into two parts in the vertical direction from the upper side at a ratio of 1: 3 are used as PUs. The size of the PU located on the upper side is 2N × (N / 2) pixels. On the other hand, the size of the PU located on the lower side is 2N × (3N / 2) pixels. This type is denoted as PU [2N × nU].

  (6) in FIG. 3 is a method in which two regions obtained by dividing a CU into two in the horizontal direction from the left side in a ratio of 1: 3 are used as PUs. The size of the PU located on the left side is (N / 2) × 2N pixels. On the other hand, the size of the PU located on the right side is (3N / 2) × 2N pixels. This type is denoted as PU [nL × 2N].

  (7) in FIG. 3 is a method in which two regions obtained by dividing a CU into two parts in the vertical direction from the lower side at a ratio of 1: 3 are used as PUs. The size of the PU located on the lower side is 2N × (N / 2) pixels. On the other hand, the size of the PU located on the upper side is 2N × (3N / 2) pixels. This type is denoted as PU [2N × nD].

  As described above, the sizes of various PUs depend on the size of the CU. In the case of CU depth 0, 2N = 64. In the case of CU depth 1, 2N = 32. In the case of CU depth 2, 2N = 16. For CU depth 3, 2N = 8. Therefore, considering the difference in size, there are 28 types of PUs.

As shown in FIG. 4, motion detection is performed in units of PUs. FIG. 4 is a diagram for explaining motion detection based on the PU. For example, the frame f corresponding to the time T _f is set as a picture to be encoded, and the frame (f−1) corresponding to the time T _f−1 is set as a reference picture. In the motion detection process, the similarity is calculated while moving the position of the PU one pixel at a time in the reference picture, and the block with the highest similarity is extracted. Then, a motion vector for the PU is obtained.

  The similarity is calculated based on, for example, SAD (Sum of Absolute Difference) defined by the following formula (1). In the following formula (1), C (x, y) is a luminance value at the coordinates (x, y) of the frame f. R (i + x, j + y) is a luminance value at the coordinates (i + x, j + y) of the frame (f−1). A method using SSD (Sum of Squared Difference) instead of SAD is also conceivable. A method using pixel values other than the luminance value is also conceivable.

  The smaller the SAD, the higher the similarity, and the larger the SAD, the lower the similarity. Therefore, in motion detection, a block of a reference picture that minimizes the SAD is extracted. In the HEVC method, such motion detection is performed many times for each type of PU to determine a suitable PU.

(Example of division processing)
For example, determination of a suitable PU (CU division processing) can be realized based on the processing flow illustrated in FIGS. 5 and 6. FIG. 5 is a first flowchart for explaining the CU division processing. FIG. 6 is a second flowchart for explaining the CU division processing. Here, the description will be made on the assumption that a certain coding apparatus executes the processing shown in FIGS. 5 and 6.

  (S11) The encoding device specifies CU depth. For example, in the case immediately after the start of the CU division process, the encoding device specifies CU depth 0. Also, when CU depth k (k> 0) has already been specified, the encoding device specifies CU depth (k + 1).

(S12) The encoding apparatus selects one CU and sets the selected CU as a target CU (CU _O ). At this time, the encoding apparatus selects one CU from among the CUs not yet selected as a target CU.

  (S13) The encoding apparatus divides the target CU into PUs of PU [2N × 2N]. Note that the target CU and the PU of PU [2N × 2N] have the same size. Therefore, the number of PUs is one. The encoding device performs motion detection in units of PUs obtained by division, and calculates an evaluation value for each PU. As the evaluation value, for example, ME (Motion Estimation) cost is used.

  The ME cost is an estimated value of a code amount used for designating a code amount representing a motion vector or a reference image. For example, the ME cost is expressed by the following formula (2). However, Ω represents the entire set of candidate modes used when encoding a PU. SAD is the sum of absolute differences defined by the above equation (1). λ is a Lagrange undetermined multiplier. R is the total code amount when encoding is performed in mode mode. In addition, although the ME cost based on SAD is illustrated here, the ME cost based on SSE can be used as an evaluation value.

ME cost (Mode ∈ Ω) = SAD + λ · R
... (2)
(S14) The encoding apparatus divides the target CU into PUs of PU [2N × N]. In this case, the PU is divided into an upper PU and a lower PU. The encoding device performs motion detection for each of the two PUs obtained by the division, and calculates an evaluation value for each PU. Then, the encoding apparatus sets the total value of the evaluation values calculated for each PU as an evaluation value corresponding to PU [2N × N]. As the evaluation value, for example, the above ME cost is used.

  (S15) The encoding apparatus divides the target CU into PUs of PU [N × 2N]. In this case, the PU is divided into a left PU and a right PU. The encoding device performs motion detection for each of the two PUs obtained by the division, and calculates an evaluation value for each PU. Then, the encoding apparatus sets the total value of the evaluation values calculated for each PU as an evaluation value corresponding to PU [N × 2N]. As the evaluation value, for example, the above ME cost is used.

  (S16) The encoding apparatus compares the evaluation value corresponding to PU [2N × 2N], the evaluation value corresponding to PU [2N × N], and the evaluation value corresponding to PU [N × 2N]. Select the smallest division method. For example, when the evaluation value corresponding to PU [2N × 2N] is the minimum, the encoding apparatus selects PU [2N × 2N] as the division method.

  (S17) The encoding device determines whether or not the evaluation value of PU [2N × 2N] is the minimum in the process of S16. If the evaluation value of PU [2N × 2N] is the minimum, the process proceeds to S18. If the evaluation value of PU [2N × 2N] is not the minimum, the process proceeds to S22 in FIG.

  (S18) The encoding apparatus divides the target CU into PUs of PU [2N × nU]. In this case, the PU is divided into an upper 1/4 PU and a lower 3/4 PU. The encoding device performs motion detection for each of the two PUs obtained by the division, and calculates an evaluation value for each PU. Then, the encoding apparatus sets the total value of the evaluation values calculated for each PU as an evaluation value corresponding to PU [2N × nU]. As the evaluation value, for example, the above ME cost is used.

  (S19) The encoding apparatus divides the target CU into PUs of PU [2N × nD]. In this case, the PU is divided into an upper 3/4 PU and a lower 1/4 PU. The encoding device performs motion detection for each of the two PUs obtained by the division, and calculates an evaluation value for each PU. Then, the encoding apparatus sets the total value of the evaluation values calculated for each PU as an evaluation value corresponding to PU [2N × nD]. As the evaluation value, for example, the above ME cost is used.

  (S20) The encoding apparatus divides the target CU into PUs of PU [nL × 2N]. In this case, the PU is divided into a left 1/4 PU and a right 3/4 PU. The encoding device performs motion detection for each of the two PUs obtained by the division, and calculates an evaluation value for each PU. Then, the encoding apparatus sets the total value of the evaluation values calculated for each PU as an evaluation value corresponding to PU [nL × 2N]. As the evaluation value, for example, the above ME cost is used. When the process of S20 is completed, the process proceeds to S21 of FIG.

  (S21) The encoding apparatus divides the target CU into PUs of PU [nR × 2N]. In this case, the PU is divided into a left 3/4 PU and a right 1/4 PU. The encoding device performs motion detection for each of the two PUs obtained by the division, and calculates an evaluation value for each PU. Then, the encoding apparatus sets the total value of the evaluation values calculated for each PU as an evaluation value corresponding to PU [nR × 2N]. As the evaluation value, for example, the above ME cost is used. When the process of S21 is completed, the process proceeds to S27.

  (S22) The encoding device determines whether or not the evaluation value of PU [2N × N] is minimum in the process of S16. If the evaluation value of PU [2N × N] is the minimum, the process proceeds to S23. If the evaluation value of PU [2N × N] is not the minimum, the process proceeds to S25.

  (S23) The encoding apparatus divides the target CU into PUs of PU [2N × nU]. In this case, the PU is divided into an upper 1/4 PU and a lower 3/4 PU. The encoding device performs motion detection for each of the two PUs obtained by the division, and calculates an evaluation value for each PU. Then, the encoding apparatus sets the total value of the evaluation values calculated for each PU as an evaluation value corresponding to PU [2N × nU]. As the evaluation value, for example, the above ME cost is used.

  (S24) The encoding apparatus divides the target CU into PUs of PU [2N × nD]. In this case, the PU is divided into an upper 3/4 PU and a lower 1/4 PU. The encoding device performs motion detection for each of the two PUs obtained by the division, and calculates an evaluation value for each PU. Then, the encoding apparatus sets the total value of the evaluation values calculated for each PU as an evaluation value corresponding to PU [2N × nD]. As the evaluation value, for example, the above ME cost is used.

  (S25) The encoding apparatus divides the target CU into PUs of PU [nL × 2N]. In this case, the PU is divided into a left 1/4 PU and a right 3/4 PU. The encoding device performs motion detection for each of the two PUs obtained by the division, and calculates an evaluation value for each PU. Then, the encoding apparatus sets the total value of the evaluation values calculated for each PU as an evaluation value corresponding to PU [nL × 2N]. As the evaluation value, for example, the above ME cost is used.

  (S26) The encoding apparatus divides the target CU into PUs of PU [nR × 2N]. In this case, the PU is divided into a left 3/4 PU and a right 1/4 PU. The encoding device performs motion detection for each of the two PUs obtained by the division, and calculates an evaluation value for each PU. Then, the encoding apparatus sets the total value of the evaluation values calculated for each PU as an evaluation value corresponding to PU [nR × 2N]. As the evaluation value, for example, the above ME cost is used.

  (S27) The encoding apparatus selects a division method corresponding to the smallest evaluation value among the calculated evaluation values, and stores the evaluation value. For example, consider the case where it is determined in the process of S16 that the evaluation value of PU [2N × 2N] is minimum. In this case, the encoding apparatus performs evaluation value PU [2N × 2N], evaluation value PU [2N × nU], evaluation value PU [2N × nD], evaluation value PU [nL × 2N], PU [ nR × 2N] are compared, and a division method corresponding to the minimum evaluation value is selected.

  Also, when it is determined in the process of S16 that the evaluation value of PU [2N × N] is the minimum, the encoding device determines the evaluation value of PU [2N × N], the evaluation value of PU [2N × nU], PU [2N × nD] evaluation values are compared, and a division method corresponding to the minimum evaluation value is selected. Also, when it is determined in the process of S16 that the evaluation value of PU [N × 2N] is the minimum, the encoding device determines the evaluation value of PU [N × 2N], the evaluation value of PU [nL × 2N], PU The evaluation values of [nR × 2N] are compared, and a division method corresponding to the minimum evaluation value is selected.

  (S28) The encoding device determines whether all CUs have been selected. If all CUs have been selected for the specified CU depth, the process proceeds to S29. On the other hand, if there is a CU that has not been selected for the specified CU depth, the process proceeds to S12 of FIG.

  (S29) The encoding device determines whether all CU depths have been specified. If all CU depths from CU depth 0 to CU depth 3 have been specified, the process proceeds to S30. On the other hand, if there is a CU depth that has not been specified, the process proceeds to S11 of FIG.

  (S30) For each CU depth, the encoding apparatus calculates a total evaluation value obtained by summing the stored evaluation values, and selects a CU depth that minimizes the total evaluation value. When the process of S30 is completed, the series of processes shown in FIGS. 5 and 6 ends.

  The processing flow illustrated in FIGS. 5 and 6 is an example of a method for realizing CU division. In the following description, a CU dividing method that is more efficient than the CU dividing method shown in this processing flow, and functions of the information processing apparatus 100 that can realize the dividing method will be described.

The CU dividing method has been described above.
[2-2. hardware]
Here, with reference to FIG. 7, hardware capable of realizing the functions of the information processing apparatus 100 described later will be described. The information processing apparatus 100 is an example of an information processing apparatus according to the second embodiment. FIG. 7 is a diagram illustrating an example of hardware capable of realizing the functions of the information processing apparatus according to the second embodiment.

  The functions of the information processing apparatus 100 can be realized using, for example, hardware resources shown in FIG. That is, the functions of the information processing apparatus 100 are realized by controlling the hardware shown in FIG. 7 using a computer program.

  As shown in FIG. 7, this hardware mainly includes a CPU 902, a ROM (Read Only Memory) 904, a RAM 906, a host bus 908, and a bridge 910. Further, this hardware includes an external bus 912, an interface 914, an input unit 916, an output unit 918, a storage unit 920, a drive 922, a connection port 924, and a communication unit 926.

  The CPU 902 functions as, for example, an arithmetic processing unit or a control unit, and controls the overall operation of each component or a part thereof based on various programs recorded in the ROM 904, the RAM 906, the storage unit 920, or the removable recording medium 928. . The ROM 904 is an example of a storage device that stores a program read by the CPU 902, data used for calculation, and the like. The RAM 906 temporarily or permanently stores, for example, a program read by the CPU 902 and various parameters that change when the program is executed.

  These elements are connected to each other via, for example, a host bus 908 capable of high-speed data transmission. On the other hand, the host bus 908 is connected to an external bus 912 having a relatively low data transmission speed via a bridge 910, for example. As the input unit 916, for example, a mouse, a keyboard, a touch panel, a touch pad, a button, a switch, a lever, or the like is used. Furthermore, as the input unit 916, a remote controller capable of transmitting a control signal using infrared rays or other radio waves may be used.

  As the output unit 918, for example, a display device such as a CRT (Cathode Ray Tube), an LCD (Liquid Crystal Display), a PDP (Plasma Display Panel), or an ELD (Electro-Luminescence Display) is used. As the output unit 918, an audio output device such as a speaker or headphones, or a printer may be used. In other words, the output unit 918 is a device that can output information visually or audibly.

  The storage unit 920 is a device for storing various data. As the storage unit 920, for example, a magnetic storage device such as an HDD is used. Further, as the storage unit 920, a semiconductor storage device such as an SSD (Solid State Drive) or a RAM disk, an optical storage device, a magneto-optical storage device, or the like may be used.

  The drive 922 is a device that reads information recorded on a removable recording medium 928 that is a removable recording medium or writes information on the removable recording medium 928. As the removable recording medium 928, for example, a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory is used.

  The connection port 924 is a port for connecting an external connection device 930 such as a USB (Universal Serial Bus) port, an IEEE 1394 port, a SCSI (Small Computer System Interface), an RS-232C port, or an optical audio terminal. For example, a printer or the like is used as the external connection device 930.

  The communication unit 926 is a communication device for connecting to the network 932. As the communication unit 926, for example, a communication circuit for wired or wireless LAN (Local Area Network), a communication circuit for WUSB (Wireless USB), a communication circuit or router for optical communication, an ADSL (Asymmetric Digital Subscriber Line) Communication circuits, routers, communication circuits for mobile phone networks, and the like are used. A network 932 connected to the communication unit 926 is a wired or wireless network, and includes, for example, the Internet, a LAN, a broadcast network, a satellite communication line, and the like.

The hardware of the information processing apparatus 100 has been described above. Note that the functions of the above-described encoding apparatus can also be realized using the hardware illustrated in FIG.
[2-3. Function of information processing apparatus]
Next, functions of the information processing apparatus 100 will be described with reference to FIG. FIG. 8 is a block diagram illustrating an example of functions of the information processing apparatus according to the second embodiment.

  As illustrated in FIG. 8, the information processing apparatus 100 includes a subtracter 101, an integer conversion unit 102, a quantization unit 103, and an entropy encoding unit 104. The information processing apparatus 100 includes an inverse quantization unit 105, an inverse integer conversion unit 106, an adder 107, a filter 108, and a frame memory 109. Furthermore, the information processing apparatus 100 includes an intra prediction unit 110, a difference calculation unit 111, a PU determination unit 112, a motion detection unit 113, a motion compensation unit 114, and a switching unit 115.

  Note that the functions of the subtractor 101, integer transform unit 102, quantization unit 103, entropy encoding unit 104, inverse quantization unit 105, inverse integer transform unit 106, adder 107, and filter 108 use the above-described CPU 902 and the like. Can be realized. In addition, the functions of the intra prediction unit 110, the difference calculation unit 111, the PU determination unit 112, the motion detection unit 113, the motion compensation unit 114, and the switching unit 115 can be realized using the above-described CPU 902 and the like. Further, the function of the frame memory 109 can be realized using the above-described RAM 906, the storage unit 920, and the like.

  The information processing apparatus 100 receives image data (hereinafter referred to as input image data) corresponding to a moving image frame to be encoded. Input image data is input to the subtractor 101. Further, the subtracter 101 is input with predicted image data generated by motion compensation. The subtracter 101 subtracts the predicted image data from the input image data to generate a prediction error signal. The prediction error signal is input to the integer conversion unit 102.

  The integer conversion unit 102 performs integer conversion on the prediction error signal to generate a coefficient signal. For example, the integer transform unit 102 performs DCT transform with integer precision of 4 × 4, 8 × 8, 16 × 16, and 32 × 32, and a set of DCT coefficients (coefficient signals) corresponding to the frequency components of the prediction error signal. Is generated. The coefficient signal is input to the quantization unit 103. The quantization unit 103 quantizes the coefficient signal to generate quantized data. The quantized data is input to the entropy encoding unit 104 and the inverse quantization unit 105.

  The entropy encoding unit 104 also receives output data from an intra prediction unit 110 and a motion detection unit 113 described later. The entropy encoding unit 104 generates encoded data by entropy encoding the quantized data, the output data of the intra prediction unit 110, and the output data of the motion detection unit 113 (for example, motion vector information). Note that entropy coding is a coding method in which a variable-length code is assigned according to the appearance frequency of a symbol.

  The inverse quantization unit 105 inversely quantizes the input quantized data to restore the coefficient signal. The restored coefficient signal is input to the inverse integer transform unit 106. The inverse integer transform unit 106 performs inverse integer transform on the input coefficient signal to restore the prediction error signal. The restored prediction error signal is a signal equivalent to the prediction error signal output from the subtractor 101. The prediction error signal restored by the inverse integer transform unit 106 is input to the adder 107. Further, the adder 107 receives predicted image data generated by motion compensation.

  The adder 107 adds the input prediction error signal and the prediction image data to generate a reference image signal (corresponding to a PU of the reference picture). The reference image signal is input to the filter 108 and the intra prediction unit 110. The filter 108 performs processing of a deblocking filter, SAO (Sample Adaptive Offset), and ALF (Adaptive Loop Filter) on the reference image signal to suppress the generation of block noise. The reference image signal processed by the filter 108 is stored in the frame memory 109. The intra prediction unit 110 generates predicted image data from neighboring pixels in the same image by intra prediction.

  The difference calculation unit 111 generates difference information indicating the similarity between blocks located near the boundary between the target CU and a CU adjacent to the target CU (hereinafter referred to as an adjacent CU). The difference information is input to the PU determination unit 112. The PU determination unit 112 narrows down the method of dividing the target CU (the type of PU to be used) based on the difference information. The contents of the process executed by the PU determination unit 112 will be described later. Information indicating the method of dividing the target CU narrowed down by the PU determination unit 112 (hereinafter, candidate PU information) is input to the motion detection unit 113.

  The motion detection unit 113 detects a motion vector using the input image data, the reference image signal read from the frame memory 109, and candidate PU information. The motion vector detected by the motion detection unit 113 is input to the motion compensation unit 114. Based on the input motion vector, the motion compensation unit 114 performs motion compensation on the reference image signal read from the frame memory 109 to generate predicted image data. The switching unit 115 selects the intra prediction unit 110 or the motion compensation unit 114 and inputs the predicted image data output by the selected element to the subtracter 101 and the adder 107.

(About the functions of the difference calculation unit 111 and the PU determination unit 112)
Here, the functions of the difference calculation unit 111 and the PU determination unit 112 will be further described with reference to FIGS. 9 and 10. FIG. 9 is a diagram for explaining a PU division method (vertical direction division) according to the second embodiment. FIG. 10 is a diagram for explaining a PU dividing method (horizontal division) according to the second embodiment.

(About vertical division)
First, the vertical division (PU [2N × N], PU [2N × nU], PU [2N × nD]) of the target CU will be considered with reference to FIG. Incidentally, denoted CU _O target CU, CU _U adjacent CU located above the CU _O, adjacent CU located below the CU _O and CU _D. The size of each CU is 2N × 2N.

In addition, the region occupying the lower 1/4 of CU _U is R _U , the region occupying the upper 1/4 of CU _O is R ₁ , and the 2N × (N / 2) region adjacent to the lower side of R ₁ is R Indicated as ₂ . Further, the region occupying the upper 1/4 of CU _D is R _D , the region occupying the lower 1/4 of CU _O is R ₃ , and the 2N × (N / 2) region adjacent to the upper side of R ₃ is R _4. Is written. R _U and R ₁ touch at the boundary between CU _U and CU _O. R ₃ and R _D are in contact with each other at the boundary between CU _O and CU _D.

The difference calculation unit 111 calculates the difference between R _U and R ₁ . This difference is an example of an evaluation value for evaluating the similarity between R _U and R ₁ . For example, the difference calculation unit 111, as the difference between R _U and R _1, and the luminance value Y _U of the pixels included in R _U, difference absolute value sum DY _1U between the luminance value Y ₁ of the pixels included in R ₁ (See equation (3) below). However, S ₁ represents a set collection of coordinates of all the pixels included in R _1.

Further, the difference calculation unit 111 calculates the difference between R ₁ and R ₂ . This difference is an example of an evaluation value for evaluating the similarity between R ₁ and R ₂ . For example, the difference calculation unit 111, R ₁ and as the difference between the R _2, and the luminance value Y ₁ of the pixels included in R _1, the sum of absolute differences DY ₁₂ between the luminance value Y ₂ of the pixels included in R ₂ (See equation (4) below).

Further, the difference calculation unit 111 calculates the difference between R ₃ and R _D. This difference is an example of an evaluation value for evaluating the similarity between R ₃ and R _D. For example, the difference calculation unit 111, a difference between R ₃ and R _D, the luminance value Y ₃ pixels included in R _3, sum of absolute difference DY _3D between the luminance value Y _D of pixels included in R _D (See equation (5) below). However, S ₃ represents a set collection of coordinates of all the pixels included in R _3.

Further, the difference calculation unit 111 calculates the difference between R ₃ and R ₄ . This difference is an example of an evaluation value for evaluating the similarity between R ₃ and R ₄ . For example, the difference calculation unit 111, R ₃ and the difference between R _4, and luminance values Y ₃ pixels included in R _3, the difference absolute value sum DY ₃₄ of the pixel luminance values Y ₄ that are included in R ₄ (See equation (6) below).

DY _1U , DY ₁₂ , DY _3D , and DY ₃₄ calculated by the difference calculation unit 111 are input to the PU determination unit 112. The PU determination unit 112 compares DY _1U , DY ₁₂ , DY _3D , and DY ₃₄ . The comparison results are divided into the following four cases, for example.

(Case # 1: DY _1U <DY ₁₂ and DY _3D <DY ₃₄ )
Case # 1 is a case in which R _U is more similar to R ₁ than R ₂ , and R _D is more similar to R ₃ than R ₄ . When DY _1U <DY ₁₂ and DY _3D <DY ₃₄ , the PU determination unit 112 compares DY _1U and DY _3D . In the case of DY _1U <DY _3D , the PU determination unit 112 selects PU [2N × nU] as a candidate for a division method related to vertical division. On the other hand, when DY _1U > DY _3D , the PU determination unit 112 selects PU [2N × nD] as a candidate for a division method for vertical division.

(Case # 2: DY _1U <DY ₁₂ and DY _3D > DY ₃₄ )
Case # 2 is a case in which R _U is more similar to R ₁ than R ₂ , and R ₄ is more similar to R ₃ than R _D. In the case of DY _1U <DY ₁₂ and DY _3D > DY ₃₄ , the PU determination unit 112 selects PU [2N × nU] as a candidate for a division method for vertical division.

(Case # 3: DY _1U > DY ₁₂ and DY _3D <DY ₃₄ )
Case # 3 is a case in which R ₂ is more similar to R ₁ than R _U , and R _D is more similar to R ₃ than R ₄ . When DY _1U > DY ₁₂ and DY _3D <DY ₃₄ , the PU determination unit 112 selects PU [2N × nD] as a candidate for a division method for vertical division.

(Case # 4: DY _1U > DY ₁₂ and DY _3D > DY ₃₄ )
Case # 4 is a case in which R ₂ is more similar to R ₁ than R _U , and R ₄ is more similar to R ₃ than R _D. When DY _1U > DY ₁₂ and DY _3D > DY ₃₄ , the PU determination unit 112 selects PU [2N × N] as a candidate for a division method for vertical division.

With the method described above, the division method of the target CU related to the vertical division is narrowed down to one candidate by the functions of the difference calculation unit 111 and the PU determination unit 112.
(About horizontal division)
Next, the horizontal division (PU [N × 2N], PU [nR × 2N], PU [nL × 2N]) of the target CU will be considered with reference to FIG. Incidentally, denoted CU _O target CU, CU adjacent CU located on the right side of the CU _{O R,} adjacent CU located on the left side of the CU _O and CU _L. The size of each CU is 2N × 2N.

An area occupying the left quarter of CU _R is R _R , an area occupying the right quarter of CU _O is R ₅ , and an (N / 2) × 2N area adjacent to the left side of R ₅ is R ₆ . write. Further, an area occupying the right quarter of CU _L is R _L , an area occupying the left quarter of CU _O is R ₇ , and an (N / 2) × 2N area adjacent to the right side of R ₇ is R ₈ . write. R _R and R ₅ touch at the boundary between CU _R and CU _O. R ₇ and R _L are in contact with each other at the boundary between CU _O and CU _L.

The difference calculation unit 111 calculates the difference between R _R and R ₅ . This difference is an example of an evaluation value for evaluating the similarity between R _R and R ₅ . For example, the difference calculation unit 111, R as the difference between _R and R _5, the luminance value Y _R of the pixels included in R _R, difference absolute value sum DY _5R between the luminance value Y ₅ of the pixels included in R ₅ (See equation (7) below). However, S ₅ represents a set collection of coordinates of all the pixels included in R _5.

In addition, the difference calculation unit 111 calculates the difference between R ₅ and R ₆ . This difference is an example of an evaluation value for evaluating the similarity between R ₅ and R ₆ . For example, the difference calculation unit 111, R ₅ and as the difference between R _6, and luminance values Y ₅ pixels included in R _5, sum of absolute difference DY ₅₆ between the luminance value Y ₆ of the pixels included in R ₆ (See equation (8) below).

Further, the difference calculation unit 111 calculates the difference between R ₇ and R _L. This difference is an example of an evaluation value for evaluating the similarity between R ₇ and R _L. For example, the difference calculation unit 111, as the difference between R ₇ and R _L, the luminance value Y ₇ of pixels included in R _7, difference absolute value sum DY _7L between the luminance value Y _L of pixels included in R _L (See equation (9) below). However, S ₇ represents a set collection of coordinates of all the pixels included in R _7.

Further, the difference calculation unit 111 calculates the difference between R ₇ and R ₈ . This difference is an example of an evaluation value for evaluating the similarity between R ₇ and R ₈ . For example, the difference calculation unit 111, R ₇ and as the difference between R _8, and luminance values Y ₇ of pixels included in R _7, difference absolute value sum DY ₇₈ between the luminance value Y ₈ pixels included in R ₈ (See equation (10) below).

DY _5R , DY ₅₆ , DY _7L , and DY ₇₈ calculated by the difference calculation unit 111 are input to the PU determination unit 112. The PU determination unit 112 compares DY _5R , DY ₅₆ , DY _7L , and DY ₇₈ . The comparison results are divided into the following four cases, for example.

(Case # 1: DY _5R <DY ₅₆ and DY _7L <DY ₇₈ )
Case # 1 is a case in which R _R is more similar to R ₅ than R ₆ , and R _L is more similar to R ₇ than R ₈ . When DY _5R <DY ₅₆ and DY _7L <DY ₇₈ , the PU determination unit 112 compares DY _5R and DY _7L . In the case of DY _5R <DY _7L , the PU determination unit 112 selects PU [nR × 2N] as a candidate for a division method for horizontal division. On the other hand, when DY _5R > DY _7L , the PU determination unit 112 selects PU [nL × 2N] as a candidate for a division method for horizontal division.

(Case # 2: DY _5R <DY ₅₆ and DY _7L > DY ₇₈ )
Case # 2 is a case in which R _R is more similar to R ₅ than R ₆ , and R ₈ is more similar to R ₇ than R _L. In the case of DY _5R <DY ₅₆ and DY _7L > DY ₇₈ , the PU determination unit 112 selects PU [nR × 2N] as a candidate for a division method for horizontal division.

(Case # 3: DY _5R > DY ₅₆ and DY _7L <DY ₇₈ )
Case # 3, who R ₆ than R _R are similar to R _5, and, towards the R _L than R ₈ is the case that is similar to R _7. When DY _5R > DY ₅₆ and DY _7L <DY ₇₈ , the PU determination unit 112 selects PU [nL × 2N] as a candidate for a division method related to the horizontal division.

(Case # 4: DY _5R > DY ₅₆ and DY _7L > DY ₇₈ )
Case # 4, who than R _R of R ₆ is similar to the R _5, and, towards than R _L R ₈ is the case that is similar to R _7. When DY _5R > DY ₅₆ and DY _7L > DY ₇₈ , the PU determination unit 112 selects PU [N × 2N] as a candidate for a division method for horizontal division.

  With the above method, the division method of the target CU regarding the horizontal division is narrowed down to one candidate by the functions of the difference calculation unit 111 and the PU determination unit 112. Note that PU [2N × 2N] is added to the candidates in addition to the method of dividing the target CU regarding the vertical division and the horizontal division. The PU determination unit 112 narrows down the method of dividing the target CU into one candidate for vertical division, one candidate for horizontal division, and PU [2N × 2N]. As a result, the processing load on the motion detection unit 113 is reduced by the amount of PU types reduced.

  The function of the information processing apparatus 100 has been described above. In the above description, the method of evaluating the similarity between regions using the sum of absolute differences of luminance is introduced. However, the similarity evaluation method applicable to the technique according to the second embodiment is not limited to this. . Hereinafter, the similarity evaluation method will be further described.

[2-4. Evaluation of similarity]
A method for evaluating similarity between regions will be further described with reference to FIG. FIG. 11 is a diagram for describing similarity evaluation according to the second embodiment. Here, for convenience of explanation, a method for evaluating the similarity between R ₁ and R _U as shown in FIG. 11 will be described as an example, but the same applies to a method for evaluating the similarity between other regions. It is.

(Absolute difference of average brightness)
A method for expressing the difference between R ₁ and R _U as the difference absolute value of the average luminance will be described. When this method is employed, the difference calculation unit 111 calculates an average value AY _U for the luminance values Y _U of all the pixels included in R _U. Further, the difference calculation unit 111 calculates an average value AY ₁ for the luminance value Y ₁ of the pixels included in R ₁ . Then, the difference calculation unit 111 calculates a difference absolute value DAY _1U between AY _U and AY ₁ . The same applies to other areas. The PU determination unit 112 narrows down the type of PU using the difference absolute value DAY _1U calculated by the difference calculation unit 111.

(Sum of absolute differences of color differences)
Next, it will be described a method of representing the difference absolute value sum of the color difference the difference between R ₁ and R _U. When this method is employed, the difference calculation unit 111 calculates the absolute difference between the color difference C _U (Cb, Cr) of the pixel included in R _U and the color difference C ₁ (Cb, Cr) of the pixel included in R _1. Calculate the sum DC _1U . The same applies to other areas. The PU determination unit 112 narrows down the types of PUs using the difference absolute value sum DC _1U calculated by the difference calculation unit 111.

(Absolute value of average color difference)
Next, a method of expressing the difference between R ₁ and R _U by the difference absolute value of the average color difference will be described. When this method is employed, the difference calculation unit 111 calculates an average value AC _U for the color differences C _U (Cb, Cr) of all the pixels included in R _U. Further, the difference calculation unit 111 calculates an average value AC ₁ for the color difference C ₁ (Cb, Cr) of the pixels included in R ₁ . Then, the difference calculation unit 111 calculates the absolute difference DAC _1U the AC _U and AC _1. The same applies to other areas. The PU determination unit 112 narrows down the type of PU using the difference absolute value DAC _1U calculated by the difference calculation unit 111 and the like.

(Difference absolute value of flatness)
Next, a method for expressing the difference between R ₁ and R _{U by} the difference absolute value of flatness will be described. When employing this method, the difference calculation unit 111 calculates R _U flatness F _U (see equation (11) below). Furthermore, the difference calculation unit 111 calculates the flatness F ₁ R ₁ (the following formula (12)). Then, the difference calculation unit 111 calculates a difference absolute value DF _1U between F _U and F ₁ . The same applies to other areas. The PU determination unit 112 narrows down the types of PUs using the absolute difference value DF _1U calculated by the difference calculation unit 111 and the like.

However, Y _U are luminance values of pixels included in R _U. Y ₁ is the luminance value of the pixel included in R ₁ . Further, AY _U is the mean value for the luminance values of all pixels included in R _U. AY ₁ is an average value related to the luminance values of all the pixels included in R ₁ . N _U is the total number of pixels included in R _U. N ₁ is the total number of pixels included in R ₁ . S _U represents a set in which the coordinates of all the pixels included in R _U are collected. S ₁ represents a set in which the coordinates of all the pixels included in R ₁ are collected.

The similarity evaluation method has been described above.
[2-5. Processing flow]
Next, the flow of processing according to the method for dividing the target CU will be described with reference to FIGS. FIG. 12 is a first flowchart for explaining CU division processing executed by the information processing apparatus according to the second embodiment. FIG. 13 is a second flowchart for explaining the CU division processing executed by the information processing apparatus according to the second embodiment. FIG. 14 is a third flowchart for explaining the CU division processing executed by the information processing apparatus according to the second embodiment.

  (S101) The information processing apparatus 100 specifies CU depth. For example, immediately after starting the CU division processing, the information processing apparatus 100 specifies CU depth 0. When CU depth k (k> 0) has already been specified, the information processing apparatus 100 specifies CU depth (k + 1).

(S102) The information processing apparatus 100 selects one CU and sets the selected CU as a target CU (CU _O ). At this time, the information processing apparatus 100 selects one CU from among the CUs that have not yet been selected as a target CU.

(S103) The information processing apparatus 100 refers to the adjacent CUs located above and below the target CU by the function of the difference calculation unit 111, and the target CU area and the adjacent CU area that are in contact at the boundary between the target CU and the adjacent CU. Calculate the difference between For example, the information processing apparatus 100 uses the function of the difference calculation unit 111 to determine the difference between R _U and R ₁ illustrated in FIG. 9, the difference between R ₁ and R ₂ , the difference between R ₃ and R _D , R ₃ And the difference between R ₄ and R ₄ is calculated.

(S104) The information processing apparatus 100 refers to the adjacent CUs located on the left and right of the target CU by the function of the difference calculation unit 111, and the target CU region and the adjacent CU region that are in contact at the boundary between the target CU and the adjacent CU Calculate the difference between For example, the information processing apparatus 100 uses the function of the difference calculation unit 111 to perform the difference between R _R and R ₅ illustrated in FIG. 10, the difference between R ₅ and R ₆ , the difference between R ₇ and R _L , R ₇ And the difference between R ₈ is calculated.

  (S105) With the function of the PU determination unit 112, the information processing apparatus 100 narrows down the division method candidates for vertical division and the division method candidates for horizontal division based on the differences calculated in S103 and S104. For example, the information processing apparatus 100 performs classification of the above-described cases # 1 to # 4 and selects a division method candidate for each of the vertical division and the horizontal division by the function of the PU determination unit 112. Further, the information processing apparatus 100 includes PU [2N × 2N] in the candidates for the division method by the function of the PU determination unit 112.

  (S106) The information processing apparatus 100 divides the target CU into PUs of PU [2N × 2N] by the function of the motion detection unit 113. Note that the target CU and the PU of PU [2N × 2N] have the same size. Therefore, the number of PUs is one. The information processing apparatus 100 performs motion detection in units of PUs obtained by the division by the function of the motion detection unit 113, and calculates an evaluation value for each PU. As the evaluation value, for example, the above-described ME cost is used. When the process of S106 is completed, the process proceeds to S107 of FIG.

  (S107) The information processing apparatus 100 determines whether or not PU [2N × N] is selected as a candidate for the division method. When PU [2N × N] is included in the candidates for the division method, the process proceeds to S108. On the other hand, when PU [2N × N] is not included in the candidates for the division method, the process proceeds to S109.

  (S108) The information processing apparatus 100 divides the target CU into PUs of PU [2N × N] by the function of the motion detection unit 113. In this case, the PU is divided into an upper 1/2 PU and a lower 1/2 PU. The information processing apparatus 100 performs motion detection for each of the two PUs obtained by the division by the function of the motion detection unit 113, and calculates an evaluation value for each PU. Then, the information processing apparatus 100 sets the total value of the evaluation values calculated for each PU as an evaluation value corresponding to PU [2N × N] by the function of the motion detection unit 113. As the evaluation value, for example, the above ME cost is used. When the process of S108 is completed, the process proceeds to S112 of FIG.

  (S109) The information processing apparatus 100 determines whether or not PU [2N × nU] is selected as a candidate for the division method. If PU [2N × nU] is included in the candidates for the division method, the process proceeds to S110. On the other hand, when PU [2N × nU] is not included in the candidates for the division method, the process proceeds to S111.

  (S110) The information processing apparatus 100 divides the target CU into PU [2N × nU] PUs by the function of the motion detection unit 113. In this case, the PU is divided into an upper 1/4 PU and a lower 3/4 PU. The information processing apparatus 100 performs motion detection for each of the two PUs obtained by the division by the function of the motion detection unit 113, and calculates an evaluation value for each PU. Then, the information processing apparatus 100 sets the total value of the evaluation values calculated for each PU as an evaluation value corresponding to PU [2N × nU] by the function of the motion detection unit 113. As the evaluation value, for example, the above ME cost is used. When the process of S110 is completed, the process proceeds to S112 of FIG.

  (S111) The information processing apparatus 100 divides the target CU into PUs of PU [2N × nD] by the function of the motion detection unit 113. In this case, the PU is divided into an upper 3/4 PU and a lower 1/4 PU. The information processing apparatus 100 performs motion detection for each of the two PUs obtained by the division by the function of the motion detection unit 113, and calculates an evaluation value for each PU. The information processing apparatus 100 sets the total value of the evaluation values calculated for each PU as an evaluation value corresponding to PU [2N × nD] by the function of the motion detection unit 113. As the evaluation value, for example, the above ME cost is used. When the process of S110 is completed, the process proceeds to S112 of FIG.

  (S112) The information processing apparatus 100 determines whether or not PU [N × 2N] is selected as a candidate for the division method. If PU [N × 2N] is included in the candidates for the division method, the process proceeds to S113. On the other hand, when PU [N × 2N] is not included in the candidates for the division method, the process proceeds to S114.

  (S113) The information processing apparatus 100 divides the target CU into PUs of PU [N × 2N] by the function of the motion detection unit 113. In this case, the PU is divided into a left half PU and a right half PU. The information processing apparatus 100 performs motion detection for each of the two PUs obtained by the division by the function of the motion detection unit 113, and calculates an evaluation value for each PU. Then, the information processing apparatus 100 sets the total value of the evaluation values calculated for each PU as an evaluation value corresponding to PU [N × 2N] by the function of the motion detection unit 113. As the evaluation value, for example, the above ME cost is used. When the process of S113 is completed, the process proceeds to S117.

  (S114) The information processing apparatus 100 determines whether PU [nL × 2N] is selected as a candidate for the division method. If PU [nL × 2N] is included in the candidates for the division method, the process proceeds to S115. On the other hand, when PU [nL × 2N] is not included in the candidates for the division method, the process proceeds to S116.

  (S115) The information processing apparatus 100 divides the target CU into PUs of PU [nL × 2N] by the function of the motion detection unit 113. In this case, the PU is divided into a left 1/4 PU and a right 3/4 PU. The information processing apparatus 100 performs motion detection for each of the two PUs obtained by the division by the function of the motion detection unit 113, and calculates an evaluation value for each PU. Then, the information processing apparatus 100 sets the total value of the evaluation values calculated for each PU as an evaluation value corresponding to PU [nL × 2N] by the function of the motion detection unit 113. As the evaluation value, for example, the above ME cost is used. When the process of S115 is completed, the process proceeds to S117.

  (S116) The information processing apparatus 100 divides the target CU into PUs of PU [nR × 2N] by the function of the motion detection unit 113. In this case, the PU is divided into a left 3/4 PU and a right 1/4 PU. The information processing apparatus 100 performs motion detection for each of the two PUs obtained by the division by the function of the motion detection unit 113, and calculates an evaluation value for each PU. Then, the information processing apparatus 100 sets the total value of the evaluation values calculated for each PU as an evaluation value corresponding to PU [nR × 2N] by the function of the motion detection unit 113. As the evaluation value, for example, the above ME cost is used.

  (S117) The information processing apparatus 100 compares the evaluation value corresponding to PU [2N × 2N], the evaluation value of the candidate related to vertical division, and the evaluation value of the candidate related to horizontal division by the function of the motion detection unit 113. Then, the information processing apparatus 100 selects a division method that minimizes the evaluation value by the function of the motion detection unit 113 and stores the evaluation value. For example, when the evaluation value corresponding to PU [2N × 2N] is the minimum, the information processing apparatus 100 stores the evaluation value of PU [2N × 2N] by the function of the motion detection unit 113.

  (S118) The information processing apparatus 100 determines whether all CUs have been selected. If all CUs have been selected for the specified CU depth, the process proceeds to S119. On the other hand, if there is a CU that has not been selected for the specified CU depth, the process proceeds to S102 in FIG.

  (S119) The information processing apparatus 100 determines whether all CU depths have been specified. If all CU depths from CU depth 0 to CU depth 3 have been specified, the process proceeds to S120. On the other hand, if there is a CU depth that has not been specified, the process proceeds to S101 in FIG.

  (S120) The information processing apparatus 100 calculates an evaluation value sum obtained by summing the stored evaluation values for each CU depth. Then, the information processing apparatus 100 selects a CU depth that minimizes the total evaluation value. When the process of S120 is completed, the series of processes shown in FIGS.

The processing flow related to the method for dividing the target CU has been described above.
As described above, motion detection and cost evaluation are performed by narrowing down the types of PUs based on the similarity between regions located at the boundary between the target CU and adjacent CUs, and the similarity between regions obtained by dividing the target CU. The number of times of executing this process is reduced, and the burden of the encoding process is reduced.

The second embodiment has been described above.
<3. Third Embodiment>
A third embodiment will be described. In the second embodiment described above, similarity between regions is evaluated using pixel values (luminance values, color differences, etc.) of pixels included in the comparison target region. In the third embodiment described below, a method for evaluating the similarity between regions using a DC (Direct Current) component of a region to be compared is proposed.

(Function of information processing device)
The function of the information processing apparatus 200 will be described with reference to FIG. The information processing apparatus 200 is an example of an information processing apparatus according to the third embodiment. Further, the functions of the information processing apparatus 200 can be realized using the hardware illustrated in FIG. FIG. 15 is a block diagram illustrating an example of the functions of the information processing apparatus according to the third embodiment.

  As illustrated in FIG. 15, the information processing apparatus 200 includes a subtractor 201, an integer conversion unit 202, a quantization unit 203, and an entropy encoding unit 204. Further, the information processing apparatus 200 includes an inverse quantization unit 205, an inverse integer conversion unit 206, an adder 207, a filter 208, and a frame memory 209. Furthermore, the information processing apparatus 200 includes an intra prediction unit 210, a DC difference calculation unit 211, a PU determination unit 212, a motion detection unit 213, a motion compensation unit 214, and a switching unit 215.

  Note that the functions of the subtractor 201, integer transform unit 202, quantization unit 203, entropy encoding unit 204, inverse quantization unit 205, inverse integer transform unit 206, adder 207, and filter 208 use the above-described CPU 902 and the like. Can be realized. Further, the functions of the intra prediction unit 210, the DC difference calculation unit 211, the PU determination unit 212, the motion detection unit 213, the motion compensation unit 214, and the switching unit 215 can be realized by using the above-described CPU 902 or the like. The function of the frame memory 209 can be realized using the above-described RAM 906, storage unit 920, and the like.

  The information processing apparatus 200 receives image data (input image data) corresponding to a moving image frame to be encoded. The input image data is input to the subtractor 201. Further, the subtracter 201 receives the predicted image data generated by the motion compensation. The subtracter 201 subtracts the predicted image data from the input image data to generate a prediction error signal. The prediction error signal is input to the integer conversion unit 202.

  The integer conversion unit 202 performs integer conversion on the prediction error signal to generate a coefficient signal. For example, the integer transform unit 202 performs DCT transform with integer precision of 4 × 4, 8 × 8, 16 × 16, and 32 × 32, and a set of DCT coefficients (coefficient signals) corresponding to the frequency components of the prediction error signal. Is generated. The coefficient signal is input to the quantization unit 203 and the DC difference calculation unit 211. The quantization unit 203 quantizes the coefficient signal to generate quantized data. The quantized data is input to the entropy encoding unit 204 and the inverse quantization unit 205.

  The entropy encoding unit 204 also receives output data from an intra prediction unit 210 and a motion detection unit 213, which will be described later. The entropy encoding unit 204 entropy-encodes the quantized data, the output data of the intra prediction unit 210, and the output data of the motion detection unit 213 (for example, motion vector information) to generate encoded data. Note that entropy coding is a coding method in which a variable-length code is assigned according to the appearance frequency of a symbol.

  The inverse quantization unit 205 restores the coefficient signal by inverse quantization of the input quantized data. The restored coefficient signal is input to the inverse integer transform unit 206. The inverse integer transform unit 206 performs inverse integer transform on the input coefficient signal to restore the prediction error signal. The restored prediction error signal is a signal equivalent to the prediction error signal output from the subtracter 201. The prediction error signal restored by the inverse integer conversion unit 206 is input to the adder 207. Further, the adder 207 receives predicted image data generated by motion compensation.

  The adder 207 adds the input prediction error signal and the prediction image data to generate a reference image signal (corresponding to the PU of the reference picture). The reference image signal is input to the filter 208 and the intra prediction unit 210. The filter 208 performs processing of a deblocking filter, SAO (Sample Adaptive Offset), and ALF (Adaptive Loop Filter) on the reference image signal to suppress the generation of block noise. The reference image signal processed by the filter 208 is stored in the frame memory 209. The intra prediction unit 210 generates predicted image data from neighboring pixels in the same image by intra prediction.

  The DC difference calculation unit 211 is based on the DC component of the coefficient signal input from the integer conversion unit 202, and the similarity between blocks located near the boundary between the target CU and a CU adjacent to the target CU (adjacent CU). The difference information representing is generated. The difference information is input to the PU determination unit 212. The PU determination unit 212 narrows down the target CU division method (type of PU to be used) based on the difference information. The contents of the process executed by the PU determination unit 212 will be described later. Information (candidate PU information) indicating the method of dividing the target CU narrowed down by the PU determination unit 212 is input to the motion detection unit 213.

  The motion detection unit 213 detects a motion vector using the input image data, the reference image signal read from the frame memory 209, and candidate PU information. The motion vector detected by the motion detection unit 213 is input to the motion compensation unit 214. The motion compensation unit 214 performs motion compensation on the reference image signal read from the frame memory 209 based on the input motion vector, and generates predicted image data. The switching unit 215 selects the intra prediction unit 210 or the motion compensation unit 214, and inputs the predicted image data output by the selected element to the subtracter 201 and the adder 207.

(Regarding the functions of the integer conversion unit 202, the DC difference calculation unit 211, and the PU determination unit 212)
Here, the functions of the DC difference calculation unit 211 and the PU determination unit 212 will be further described with reference to FIGS. 16 and 17. FIG. 16 is a diagram for explaining a PU division method (vertical direction division) and similarity evaluation according to the third embodiment. FIG. 17 is a diagram for explaining a PU division method (horizontal division) and similarity evaluation according to the third embodiment.

(About vertical division)
First, with reference to FIG. 16, the vertical division (PU [2N × N], PU [2N × nU], PU [2N × nD]) of the target CU will be considered. Incidentally, denoted CU _O target CU, CU _U adjacent CU located above the CU _O, adjacent CU located below the CU _O and CU _D. The size of each CU is 2N × 2N.

In addition, the region occupying the lower 1/4 of CU _U is R _U , the region occupying the upper 1/4 of CU _O is R ₁ , and the 2N × (N / 2) region adjacent to the lower side of R ₁ is R Indicated as ₂ . Further, the region occupying the upper 1/4 of CU _D is R _D , the region occupying the lower 1/4 of CU _O is R ₃ , and the 2N × (N / 2) region adjacent to the upper side of R ₃ is R _4. Is written. R _U and R ₁ touch at the boundary between CU _U and CU _O. R ₃ and R _D are in contact with each other at the boundary between CU _O and CU _D.

The integer conversion unit 202 divides R _U into four rectangular blocks B _U1 , B _U2 , B _U3 , B _U4 . In addition, the integer conversion unit 202 performs DCT conversion on the pixel values of the rectangular blocks B _U1 , B _U2 , B _U3 , and B _U4 to generate coefficient signals. Further, the integer conversion unit 202 extracts a DC component (hereinafter referred to as a DC value) from coefficient signals calculated for each of the rectangular blocks B _U1 , B _U2 , B _U3 , and B _U4 , and inputs them to the DC difference calculation unit 211. To do. The DC difference calculation unit 211 sums the DC values of the rectangular blocks B _U1 , B _U2 , B _U3 , B _{U4 and} sets the total value as the DC value V _U of R _U.

The integer conversion unit 202 divides R ₁ into four rectangular blocks B ₁₁ , B ₁₂ , B ₁₃ , and B ₁₄ . In addition, the integer conversion unit 202 performs DCT conversion on the pixel values of the rectangular blocks B ₁₁ , B ₁₂ , B ₁₃ , and B ₁₄ to generate coefficient signals. Further, the integer conversion unit 202 extracts the DC values of the rectangular blocks B ₁₁ , B ₁₂ , B ₁₃ , and B ₁₄ and inputs them to the DC difference calculation unit 211. The DC difference calculation unit 211 adds the DC values of the rectangular blocks B ₁₁ , B ₁₂ , B ₁₃ , and B _{14 and} sets the total value as the DC value V ₁ of R ₁ .

The integer conversion unit 202 divides R ₂ into four rectangular blocks B ₂₁ , B ₂₂ , B ₂₃ , and B ₂₄ . In addition, the integer conversion unit 202 performs DCT conversion on the pixel values of the rectangular blocks B ₂₁ , B ₂₂ , B ₂₃ , and B ₂₄ to generate coefficient signals. Further, the integer conversion unit 202 extracts the DC values of the rectangular blocks B ₂₁ , B ₂₂ , B ₂₃ , and B ₂₄ and inputs them to the DC difference calculation unit 211. The DC difference calculation unit 211 sums up the DC values of the rectangular blocks B ₂₁ , B ₂₂ , B ₂₃ , and B _{24 and} sets the total value as the DC value V ₂ of R ₂ .

Also, DC difference calculation unit 211, as the difference between R _U and R _1, R and DC values V _U of _U, the R ₁ DC value difference absolute value DV _1U of _{V 1 (DV 1U = | V} U - V ₁ |) is calculated. Also, DC difference calculation unit 211, R ₁ and as the difference between the R _2, a DC value V ₁ of the R _1, the difference absolute value between the DC value V ₂ of the _{R 2 DV 12 (DV 12 =} | V 1 - V ₂ |) is calculated.

The integer conversion unit 202 divides R _D into four rectangular blocks B _D1 , B _D2 , B _D3 , and B _D4 . In addition, the integer conversion unit 202 performs DCT conversion on the pixel values of the rectangular blocks B _D1 , B _D2 , B _D3 , and B _D4 to generate coefficient signals. Further, the integer conversion unit 202 extracts the DC values of the rectangular blocks B _D1 , B _D2 , B _D3 , and B _D4 and inputs them to the DC difference calculation unit 211. The DC difference calculation unit 211 sums the DC values of the rectangular blocks B _D1 , B _D2 , B _D3 , and B _{D4 and} sets the total value as the DC value V _D of R _D.

Further, the integer conversion unit 202 divides the R ₃ into four rectangular blocks _{B 31, B 32, B 33} , B 34. In addition, the integer conversion unit 202 performs DCT conversion on the pixel values of the rectangular blocks B ₃₁ , B ₃₂ , B ₃₃ , and B ₃₄ to generate coefficient signals. Further, the integer conversion unit 202 extracts the DC values of the rectangular blocks B ₃₁ , B ₃₂ , B ₃₃ , and B ₃₄ and inputs them to the DC difference calculation unit 211. The DC difference calculation unit 211 sums up the DC values of the rectangular blocks B ₃₁ , B ₃₂ , B ₃₃ , and B _{34 and} sets the total value as the DC value V ₃ of R ₃ .

The integer conversion unit 202 divides R ₄ into four rectangular blocks B ₄₁ , B ₄₂ , B ₄₃ , and B ₄₄ . In addition, the integer conversion unit 202 performs DCT conversion on the pixel values of the rectangular blocks B ₄₁ , B ₄₂ , B ₄₃ , and B ₄₄ to generate coefficient signals. Further, the integer conversion unit 202 extracts the DC values of the rectangular blocks B ₄₁ , B ₄₂ , B ₄₃ , and B ₄₄ and inputs them to the DC difference calculation unit 211. The DC difference calculation unit 211 _adds the DC values of the rectangular blocks B ₄₁ , B ₄₂ , B ₄₃ , and B ₄₄ , and sets the total value as the DC value V ₄ of R ₄ .

Also, DC difference calculation unit 211, a difference between R _D and R _3, R _D DC value V _D and the difference between the DC value V ₃ of R ₃ absolute value _{DV 3D (DV 3D = | V} D - V ₃ |) is calculated. Also, DC difference calculation unit 211, R ₃ and the difference between R _4, and DC value V ₃ of R _3, the difference absolute value between the DC value V ₄ of _{R 4 DV 34 (DV 34 =} | V 3 - V ₄ |) is calculated.

DV _1U , DV ₁₂ , DV _3D , DV ₃₄ calculated by the DC difference calculation unit 211 are input to the PU determination unit 212. The PU determination unit 212 compares DV _1U , DV ₁₂ , DV _3D , and DV ₃₄ . The comparison results are divided into the following four cases, for example.

(Case # 1: DV _1U <DV ₁₂ and DV _3D <DV ₃₄ )
Case # 1 is a case in which R _U is more similar to R ₁ than R ₂ , and R _D is more similar to R ₃ than R ₄ . When DV _1U <DV ₁₂ and DV _3D <DV ₃₄ , the PU determination unit 212 compares DV _1U with DV _3D . In the case of DV _1U <DV _3D , the PU determination unit 212 selects PU [2N × nU] as a candidate for a division method related to vertical division. On the other hand, when DV _1U > DV _3D , the PU determination unit 212 selects PU [2N × nD] as a candidate for a division method related to vertical division.

(Case # 2: DV _1U <DV ₁₂ and DV _3D > DV ₃₄ )
Case # 2 is a case in which R _U is more similar to R ₁ than R ₂ , and R ₄ is more similar to R ₃ than R _D. When DV _1U <DV ₁₂ and DV _3D > DV ₃₄ , the PU determination unit 212 selects PU [2N × nU] as a candidate for a division method related to vertical division.

(Case # 3: DV _1U > DV ₁₂ and DV _3D <DV ₃₄ )
Case # 3 is a case in which R ₂ is more similar to R ₁ than R _U , and R _D is more similar to R ₃ than R ₄ . When DV _1U > DV ₁₂ and DV _3D <DV ₃₄ , the PU determination unit 212 selects PU [2N × nD] as a candidate for a division method related to vertical division.

(Case # 4: DV _1U > DV ₁₂ and DV _3D > DV ₃₄ )
Case # 4 is a case in which R ₂ is more similar to R ₁ than R _U , and R ₄ is more similar to R ₃ than R _D. When DV _1U > DV ₁₂ and DV _3D > DV ₃₄ , the PU determination unit 212 selects PU [2N × N] as a candidate for a division method related to vertical division.

With the above method, the method of dividing the target CU regarding the vertical division is narrowed down to one candidate.
(About horizontal division)
Next, the vertical division (PU [N × 2N], PU [nR × 2N], PU [nL × 2N]) of the target CU will be considered with reference to FIG. Incidentally, denoted CU _O target CU, CU adjacent CU located on the right side of the CU _{O R,} adjacent CU located on the left side of the CU _O and CU _L. The size of each CU is 2N × 2N.

An area occupying the left quarter of CU _R is R _R , an area occupying the right quarter of CU _O is R ₅ , and an (N / 2) × 2N area adjacent to the left side of R ₅ is R ₆ . write. Further, an area occupying the right quarter of CU _L is R _L , an area occupying the left quarter of CU _O is R ₇ , and an (N / 2) × 2N area adjacent to the right side of R ₇ is R ₈ . write. R _R and R ₅ touch at the boundary between CU _R and CU _O. R ₇ and R _L are in contact with each other at the boundary between CU _O and CU _L.

The integer conversion unit 202 divides R _R into four rectangular blocks B _R1 , B _R2 , B _R3 , and B _R4 . In addition, the integer conversion unit 202 performs DCT conversion on the pixel values of the rectangular blocks B _R1 , B _R2 , B _R3 , and B _R4 to generate coefficient signals. Further, the integer conversion unit 202 extracts the DC values of the rectangular blocks B _R1 , B _R2 , B _R3 , and B _R4 and inputs them to the DC difference calculation unit 211. The DC difference calculation unit 211 sums the DC values of the rectangular blocks B _R1 , B _R2 , B _R3 , B _{R4 and} sets the total value as the DC value V _R of R _R.

The integer conversion unit 202 divides R ₅ into four rectangular blocks B ₅₁ , B ₅₂ , B ₅₃ , and B ₅₄ . In addition, the integer conversion unit 202 performs DCT conversion on the pixel values of the rectangular blocks B ₅₁ , B ₅₂ , B ₅₃ , and B ₅₄ to generate coefficient signals. Further, the integer conversion unit 202 extracts the DC values of the rectangular blocks B ₅₁ , B ₅₂ , B ₅₃ , and B ₅₄ and inputs them to the DC difference calculation unit 211. The DC difference calculation unit 211 sums up the DC values of the rectangular blocks B ₅₁ , B ₅₂ , B ₅₃ , and B ₅₄ , and sets the total value as the DC value V ₅ of R ₅ .

The integer conversion unit 202 divides R ₆ into four rectangular blocks B ₆₁ , B ₆₂ , B ₆₃ , and B ₆₄ . The integer conversion unit 202 DCT converts the pixel values of the rectangular blocks B ₆₁ , B ₆₂ , B ₆₃ , and B ₆₄ , respectively, to generate coefficient signals. Further, the integer conversion unit 202 extracts the DC values of the rectangular blocks B ₆₁ , B ₆₂ , B ₆₃ , and B ₆₄ and inputs them to the DC difference calculation unit 211. The DC difference calculation unit 211 _adds the DC values of the rectangular blocks B ₆₁ , B ₆₂ , B ₆₃ , and B ₆₄ , and sets the total value as the DC value V ₆ of R ₆ .

Also, DC difference calculation unit 211, R as the difference between _R and R _5, R a DC value V _R of the _R, the difference absolute value DV _5R the DC value V ₅ of _{R 5 (DV 5R = | V} R - V ₅ |) is calculated. Also, DC difference calculation unit 211, as the difference between R ₅ and R _6, R a DC value V ₅ of _5, absolute difference between the DC value V ₆ of _{R 6 DV 56 (DV 56 =} | V 5 - V ₆ |) is calculated.

The integer conversion unit 202 divides R _L into four rectangular blocks B _L1 , B _L2 , B _L3 , and B _L4 . In addition, the integer conversion unit 202 performs DCT conversion on the pixel values of the rectangular blocks B _L1 , B _L2 , B _L3 , and B _L4 to generate coefficient signals. Further, the integer conversion unit 202 extracts the DC values of the rectangular blocks B _L1 , B _L2 , B _L3 , and B _L4 and inputs them to the DC difference calculation unit 211. The DC difference calculation unit 211 _adds the DC values of the rectangular blocks B _L1 , B _L2 , B _L3 , and B _{L4 and} sets the total value as the DC value V _L of R _L.

The integer conversion unit 202 divides R ₇ into four rectangular blocks B ₇₁ , B ₇₂ , B ₇₃ , and B ₇₄ . The integer conversion unit 202 DCT converts the pixel values of the rectangular blocks B ₇₁ , B ₇₂ , B ₇₃ , and B ₇₄ , respectively, to generate coefficient signals. Further, the integer conversion unit 202 extracts the DC values of the rectangular blocks B ₇₁ , B ₇₂ , B ₇₃ , and B ₇₄ and inputs them to the DC difference calculation unit 211. The DC difference calculation unit 211 sums the DC values of the rectangular blocks B ₇₁ , B ₇₂ , B ₇₃ , and B ₇₄ , and sets the total value as the DC value V ₇ of R ₇ .

The integer conversion unit 202 divides R ₈ into four rectangular blocks B ₈₁ , B ₈₂ , B ₈₃ , and B ₈₄ . In addition, the integer conversion unit 202 performs DCT conversion on the pixel values of the rectangular blocks B ₈₁ , B ₈₂ , B ₈₃ , and B ₈₄ to generate coefficient signals. Further, the integer conversion unit 202 extracts the DC values of the rectangular blocks B ₈₁ , B ₈₂ , B ₈₃ , and B ₈₄ and inputs them to the DC difference calculation unit 211. The DC difference calculation unit 211 _{adds up} the DC values of the rectangular blocks B ₈₁ , B ₈₂ , B ₈₃ , and B _{84 and} sets the total value as the DC value V ₈ of R ₈ .

Also, DC difference calculation unit 211, as the difference between the R _L and R _7, R a DC value V _L of the _L, R ₇ DC value V difference absolute value DV _7L and _{7 (DV 7L = | V L} - V ₇ |) is calculated. Also, DC difference calculation unit 211, R ₇ and as the difference between R _8, a DC value V ₇ of R _7, a difference absolute value between the DC value V ₈ of _{R 8 DV 78 (DV 78 =} | V 7 - V ₈ |) is calculated.

DV _5R , DV ₅₆ , DV _7L , DV ₇₈ calculated by the DC difference calculation unit 211 are input to the PU determination unit 212. The PU determination unit 212 compares DV _5R , DV ₅₆ , DV _7L , and DV ₇₈ . The comparison results are divided into the following four cases, for example.

(Case # 1: DV _5R <DV ₅₆ and DV _7L <DV ₇₈ )
Case # 1 is a case in which R _R is more similar to R ₅ than R ₆ , and R _L is more similar to R ₇ than R ₈ . When DV _5R <DV ₅₆ and DV _7L <DV ₇₈ , the PU determination unit 212 compares DV _5R with DV _7L . When DV _5R <DV _7L , the PU determination unit 212 selects PU [nR × 2N] as a candidate for a division method related to the horizontal division. On the other hand, when DV _5R > DV _7L , the PU determination unit 212 selects PU [nL × 2N] as a candidate for a division method related to the horizontal division.

(Case # 2: DV _5R <DV ₅₆ and DV _7L > DV ₇₈ )
Case # 2 is a case in which R _R is more similar to R ₅ than R ₆ , and R ₈ is more similar to R ₇ than R _L. In the case of DV _5R <DV ₅₆ and DV _7L > DV ₇₈ , the PU determination unit 212 selects PU [nR × 2N] as a candidate for a division method related to horizontal division.

(Case # 3: DV _5R > DV ₅₆ and DV _7L <DV ₇₈ )
Case # 3, who R ₆ than R _R are similar to R _5, and, towards the R _L than R ₈ is the case that is similar to R _7. In the case of DV _5R > DV ₅₆ and DV _7L <DV ₇₈ , the PU determination unit 212 selects PU [nL × 2N] as a candidate for a division method for horizontal division.

(Case # 4: DV _5R > DV ₅₆ and DV _7L > DV ₇₈ )
Case # 4, who than R _R of R ₆ is similar to the R _5, and, towards than R _L R ₈ is the case that is similar to R _7. When DV _5R > DV ₅₆ and DV _7L > DV ₇₈ , the PU determination unit 212 selects PU [N × 2N] as a candidate for a division method for horizontal division.

  By the above method, the method of dividing the target CU regarding the horizontal division is narrowed down to one candidate. Note that PU [2N × 2N] is added to the candidates in addition to the method of dividing the target CU regarding the vertical division and the horizontal division. The PU determination unit 212 narrows down the method of dividing the target CU into one candidate for vertical division, one candidate for horizontal division, and PU [2N × 2N]. As a result, the processing load on the motion detection unit 213 is reduced by the amount that the type of PU is reduced.

  The function of the information processing apparatus 200 has been described above. In the above description, the method of using the total value obtained by summing the DC values of all blocks for each region as the DC value of each region is exemplified. For example, instead of the sum of DC values, the average value of DC values is used. It is possible to make modifications such as A method for evaluating the similarity between regions using the sum of absolute differences of DC values calculated for each block is also conceivable. Such modifications also belong to the technical scope of the third embodiment.

(About processing flow)
Here, the process of dividing the target CU according to the third embodiment will be described. The target CU division processing flow according to the third embodiment is obtained by modifying the processing of S103, S104, and S105 into processing based on a DC value in the processing flow according to the second embodiment shown in FIGS. It becomes. That is, the method for evaluating the similarity between regions is modified to the method described above. Since other processes are substantially the same, detailed description thereof is omitted.

The third embodiment has been described above.
As mentioned above, although preferred embodiment was described referring an accompanying drawing, this invention is not limited to the example which concerns. It is obvious for a person skilled in the art that various variations and modifications can be conceived within the scope of the claims, and such variations and modifications are naturally understood by the technical scope of the present invention. It goes without saying that it belongs to a range.

<4. Addendum>
The following additional notes are disclosed with respect to the embodiment described above.
(Supplementary note 1) A storage unit that stores a target block and an adjacent block adjacent to the target block;
The similarity between the first area that touches the boundary with the adjacent block in the target block and the adjacent area that touches the boundary in the adjacent block is evaluated, and the first area and the target block To evaluate the similarity with the second region in contact with the first region,
When the similarity between the first region and the adjacent region is higher than the similarity between the first region and the second region, the first region is selected from the first region and the target block. And an arithmetic unit that uses the excluded area as a candidate for a predictive coding unit.

(Additional remark 2) When the similarity of the said 1st area | region and the said adjacent area is lower than the similarity of the said 1st area | region and the said 2nd area | region, the said calculating part is said 1st area | region and said 2nd area | region. The information processing device according to attachment 1, wherein a third region combined with a region and a region obtained by removing the third region from the target block are candidates for predictive coding units.

(Supplementary Note 3) The first region is a region that occupies 1/4 on the boundary side in the target block,
The information processing apparatus according to appendix 1 or 2, wherein the second area and the adjacent area are areas having the same size as the first area.

(Additional remark 4) The said memory | storage part adjoins the said target block toward the 1st adjacent block adjacent to the said target block toward a 1st direction, and the 2nd direction orthogonal to the said 1st direction. Storing the second adjacent block;
The arithmetic unit evaluates the similarity between the first area and the adjacent area and the similarity between the first area and the second area for each of the first adjacent block and the second adjacent block. The information processing apparatus according to any one of appendices 1 to 3, wherein a candidate for the predictive coding unit is determined.

(Additional remark 5) The said calculating part evaluates the similarity of the said 1st area | region and the said adjacent area based on the difference of the brightness | luminance of the pixel contained in the said 1st area | region, and the brightness | luminance of the pixel contained in the said adjacent area. The similarity between the first region and the second region is evaluated based on the difference between the luminance of the pixel included in the first region and the luminance of the pixel included in the second region. An information processing apparatus according to claim 1.

(Additional remark 6) The said calculating part evaluates the similarity of the said 1st area | region and the said adjacent area based on the difference of the color difference of the pixel contained in the said 1st area | region, and the color difference of the pixel contained in the said adjacent area | region. The similarity between the first region and the second region is evaluated based on the difference between the luminance of the pixel included in the first region and the luminance of the pixel included in the second region. An information processing apparatus according to claim 1.

(Additional remark 7) The said calculating part evaluates the similarity of the said 1st area | region and the said adjacent area based on the difference of the flatness of the said 1st area | region, and the flatness of the said adjacent area, The information processing apparatus according to any one of appendices 1 to 4, wherein the similarity between the first region and the second region is evaluated based on a difference between a flatness and a flatness of the second region.

(Additional remark 8) The said calculating part evaluates the similarity of the said 1st area | region and the said adjacent area based on the difference of the value of the DC component which the said 1st area | region has, and the value of the DC component which the said adjacent area | region has. The similarity between the first region and the second region is evaluated based on the difference between the direct current component value of the first region and the direct current component value of the second region. An information processing apparatus according to claim 1.

(Supplementary Note 9) The storage unit stores an upper adjacent block adjacent to the upper side of the target block and a lower adjacent block adjacent to the lower side of the target block,
The computing unit is
For each of the upper adjacent block and the lower adjacent block, the similarity between the first region and the adjacent region and the similarity between the first region and the second region are evaluated, and the first region is evaluated. Extract the region with the highest similarity to
When the first region that is in contact with the upper adjacent block and the adjacent region of the upper adjacent block are extracted as the set of regions having the highest similarity, the first region and the target block are extracted from the first region and the target block. The region excluding the first region is set as a candidate for the predictive coding unit,
When the first region that is in contact with the lower adjacent block and the adjacent region of the lower adjacent block are extracted as the set of regions having the highest similarity, the first region and the target block A region excluding the first region from among the candidates for the predictive coding unit,
When the first region and the second region are extracted as the set of regions having the highest similarity, the third region obtained by combining the first region and the second region, and the target block The information processing apparatus according to attachment 1, wherein an area excluding the third area is a candidate for the predictive coding unit.

(Supplementary Note 10) The storage unit stores a left adjacent block adjacent to the left side of the target block and a right adjacent block adjacent to the right side of the target block,
The computing unit is
For each of the left adjacent block and the right adjacent block, the similarity between the first region and the adjacent region and the similarity between the first region and the second region are evaluated, and the first region and Extract the region with the highest similarity
When the first region that touches the left adjacent block and the adjacent region of the left adjacent block are extracted as the set of regions having the highest similarity, the first region and the target block are extracted from the first region and the target block. The region excluding the first region is set as a candidate for the predictive coding unit,
When the first region that contacts the right adjacent block and the adjacent region of the right adjacent block are extracted as the set of regions having the highest similarity, the first region and the target block are extracted from the first region and the target block. The region excluding the first region is set as a candidate for the predictive coding unit,
When the first region and the second region are extracted as the set of regions having the highest similarity, the third region obtained by combining the first region and the second region, and the target block The information processing apparatus according to appendix 1 or 9, wherein an area excluding the third area is a candidate for the predictive coding unit.

(Additional remark 11) The computer which can read the information on the said target block and the said adjacent block from the memory | storage part which memorize | stores a target block and the adjacent block adjacent to the said target block,
The similarity between the first area that touches the boundary with the adjacent block in the target block and the adjacent area that touches the boundary in the adjacent block is evaluated, and the first area and the target block To evaluate the similarity with the second region in contact with the first region,
When the similarity between the first region and the adjacent region is higher than the similarity between the first region and the second region, the first region is selected from the first region and the target block. A coding unit selection method in which the excluded area is a candidate for a predictive coding unit.

(Supplementary note 12) The computer
A third region obtained by combining the first region and the second region when the similarity between the first region and the adjacent region is lower than the similarity between the first region and the second region; The method for selecting a coding unit according to attachment 11, wherein a region obtained by removing the third region from the target block is a candidate for a prediction coding unit.

(Supplementary Note 13) The first area is an area that occupies 1/4 of the target block on the boundary side,
The method for selecting a coding unit according to attachment 11 or 12, wherein the second area and the adjacent area are areas having the same size as the first area.

(Additional remark 14) The said memory | storage part adjoins the said target block toward the 1st adjacent block adjacent to the said target block toward a 1st direction, and the 2nd direction orthogonal to the said 1st direction. Storing the second adjacent block;
The computer is
For each of the first adjacent block and the second adjacent block, the similarity between the first region and the adjacent region and the similarity between the first region and the second region are evaluated, and the prediction code A method for selecting a coding unit according to any one of appendices 11 to 13, wherein a candidate for a coding unit is determined.

(Supplementary note 15)
Pixels included in the first region are evaluated based on the difference between the luminance of the pixels included in the first region and the luminance of the pixels included in the adjacent region, and the similarity between the first region and the adjacent region is evaluated. The similarity between the first area and the second area is evaluated based on the difference between the luminance of the first area and the luminance of the pixels included in the second area. Method.

(Supplementary Note 16) The computer
Pixels included in the first region are evaluated based on a difference between a color difference between pixels included in the first region and a color difference between pixels included in the adjacent region, and the similarity between the first region and the adjacent region is evaluated. The similarity between the first area and the second area is evaluated based on the difference between the luminance of the first area and the luminance of the pixels included in the second area. Method.

(Supplementary Note 17)
Based on the difference between the flatness of the first region and the flatness of the adjacent region, the similarity between the first region and the adjacent region is evaluated, and the flatness of the first region and the flatness of the second region are evaluated. The method for selecting a coding unit according to any one of appendices 11 to 14, wherein the similarity between the first region and the second region is evaluated based on a difference in degree.

(Supplementary note 18)
The similarity between the first region and the adjacent region is evaluated based on the difference between the direct current component value of the first region and the direct current component value of the adjacent region, and the direct current component of the first region The selection of the coding unit according to any one of appendices 11 to 14, wherein the similarity between the first region and the second region is evaluated based on the difference between the value of the DC component and the value of the DC component of the second region Method.

(Additional remark 19) From the memory | storage part which memorize | stores a target block and the adjacent block adjacent to the said target block, to the computer which can read the information of the said target block and the said adjacent block,
The similarity between the first area that touches the boundary with the adjacent block in the target block and the adjacent area that touches the boundary in the adjacent block is evaluated, and the first area and the target block To evaluate the similarity with the second region in contact with the first region,
When the similarity between the first region and the adjacent region is higher than the similarity between the first region and the second region, the first region is selected from the first region and the target block. A program that executes a process in which the excluded area is a candidate for a predictive coding unit.

(Additional remark 20) From the memory | storage part which memorize | stores a target block and the adjacent block adjacent to the said target block, to the computer which can read the information of the said target block and the said adjacent block,
The similarity between the first area that touches the boundary with the adjacent block in the target block and the adjacent area that touches the boundary in the adjacent block is evaluated, and the first area and the target block To evaluate the similarity with the second region in contact with the first region,
When the similarity between the first region and the adjacent region is higher than the similarity between the first region and the second region, the first region is selected from the first region and the target block. A computer-readable recording medium on which a program for executing processing for setting the excluded area as a candidate for a predictive coding unit is recorded.

DESCRIPTION OF SYMBOLS 10 Information processing apparatus 11 Memory | storage part 12 Calculation part A1 1st area A2 2nd area A3 3rd area A4 Area except 3rd area A5 Area except 1st area An Adjacent area BLo Target block BLn Adjacent block EV12, EV1n Evaluation

Claims (10)

A storage unit for storing the target block and an adjacent block adjacent to the target block;
The similarity between the first area that touches the boundary with the adjacent block in the target block and the adjacent area that touches the boundary in the adjacent block is evaluated, and the first area and the target block To evaluate the similarity with the second region in contact with the first region,
When the similarity between the first region and the adjacent region is higher than the similarity between the first region and the second region, the first region is selected from the first region and the target block. And an arithmetic unit that uses the excluded area as a candidate for a predictive coding unit. The arithmetic unit combines the first region and the second region when the similarity between the first region and the adjacent region is lower than the similarity between the first region and the second region. The information processing apparatus according to claim 1, wherein the third area and the area obtained by removing the third area from the target block are candidates for predictive coding units. The first area is an area that occupies 1/4 of the target block on the boundary side,
The information processing apparatus according to claim 1, wherein the second area and the adjacent area are areas having the same size as the first area. The storage unit includes a first adjacent block adjacent to the target block in a first direction and a second adjacent adjacent to the target block in a second direction orthogonal to the first direction. Remember block and
The arithmetic unit evaluates the similarity between the first area and the adjacent area and the similarity between the first area and the second area for each of the first adjacent block and the second adjacent block. The information processing apparatus according to claim 1, wherein a candidate for the predictive coding unit is determined. The computing unit evaluates similarity between the first region and the adjacent region based on a difference between a luminance of a pixel included in the first region and a luminance of a pixel included in the adjacent region. The similarity between the first region and the second region is evaluated based on the difference between the luminance of the pixel included in the region and the luminance of the pixel included in the second region. The information processing apparatus described in 1. The computing unit evaluates the similarity between the first region and the adjacent region based on a difference between a color difference between pixels included in the first region and a color difference between pixels included in the adjacent region, The similarity between the first region and the second region is evaluated based on the difference between the luminance of the pixel included in the region and the luminance of the pixel included in the second region. The information processing apparatus described in 1. The computing unit evaluates the similarity between the first region and the adjacent region based on the difference between the flatness of the first region and the flatness of the adjacent region, and the flatness of the first region and the flatness of the first region The information processing apparatus according to claim 1, wherein the similarity between the first area and the second area is evaluated based on a difference from the flatness of the second area. The computing unit evaluates the similarity between the first region and the adjacent region based on a difference between a direct current component value of the first region and a direct current component value of the adjacent region. The similarity between the first region and the second region is evaluated based on a difference between a DC component value of the region and a DC component value of the second region. The information processing apparatus described in 1. A computer that can read out information on the target block and the adjacent block from a storage unit that stores the target block and an adjacent block adjacent to the target block.
The similarity between the first area that touches the boundary with the adjacent block in the target block and the adjacent area that touches the boundary in the adjacent block is evaluated, and the first area and the target block To evaluate the similarity with the second region in contact with the first region,
When the similarity between the first region and the adjacent region is higher than the similarity between the first region and the second region, the first region is selected from the first region and the target block. A coding unit selection method in which the excluded area is a candidate for a predictive coding unit. From the storage unit that stores the target block and the adjacent block adjacent to the target block, to the computer that can read the information of the target block and the adjacent block,
The similarity between the first area that touches the boundary with the adjacent block in the target block and the adjacent area that touches the boundary in the adjacent block is evaluated, and the first area and the target block To evaluate the similarity with the second region in contact with the first region,
When the similarity between the first region and the adjacent region is higher than the similarity between the first region and the second region, the first region is selected from the first region and the target block. A program that executes a process in which the excluded area is a candidate for a predictive coding unit.

Images