JP2024002320A - Image encoding device, method, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image encoding device capable of suppressing an amount of encoding processing while securing the quality.

SOLUTION: In an image encoding device encoding an image by applying a prediction in units of block, when an encoding target block, to which an inter-prediction and a geometric partitioning are applied (111) as a prediction mode, is used, encoding target block information that is information on the encoding target block is extracted (1112), and, based on the encoding target block information, it is determined that only a partial set selected (1113) from the entire of a geometric partitioning mode set that is defined in the geometric partitioning in advance is an evaluation target (1114) for use in encoding after an inter-prediction and a geometric partitioning are applied to the encoding target block.

SELECTED DRAWING: Figure 3

COPYRIGHT: (C)2024,JPO&INPIT

Description

The present invention relates to an image encoding device, method, and program that apply geometric division.

In Non-Patent Document 1, regarding a technique called geometric segmentation, an encoding device divides an encoded block using a predefined method, and predicts each region using motion information generated by a merging unit. A technique for encoding is disclosed. Here, motion information refers to a motion vector and an index that specifies a reference frame.

Algorithm description for Versatile Video Coding and Test Model 11 (VTM 11), JVET-T2002.

The geometric division in Non-Patent Document 1 has a problem in that it requires a large amount of encoding processing.

That is, in Non-Patent Document 1, one geometric division mode to be applied to the target encoding block is determined from the geometric division mode set. Here, the geometric division mode is information composed of a division boundary (a line that divides the target encoding block into two regions) and two pieces of motion information used for predictive encoding. Moreover, the geometric division mode set is a set whose elements are geometric division modes. In order to select the geometric partitioning mode that minimizes the rate-distortion cost (cost by the cost function J=D+λR etc. due to bit rate R and distortion D) from the geometric partitioning mode set, the geometric partitioning mode set is When the number of elements is large, there is a problem that a very large amount of encoding processing is required.

In view of the above problems of the prior art, an object of the present invention is to provide an image encoding device, method, and program that can suppress the amount of encoding processing while ensuring quality.

In order to achieve the above object, the present invention provides an image encoding device that encodes a video by applying prediction in units of blocks. When handling the encoding target block, the encoding target block information, which is information about the encoding target block, is extracted, and based on the encoding target block information, the geometric division mode set predefined in the geometric division is set. The first feature is that only a partial set selected from the entire block is used as an evaluation target for encoding after applying inter prediction and geometric division to the encoding target block. In addition, in an image encoding device that encodes video by applying prediction in units of blocks, when dealing with a block to be encoded to which inter prediction and geometric division are applied as prediction modes, the geometric Only a partial set selected from the entire geometric division mode set predefined in the division is used for encoding after applying inter prediction and geometric division to the block to be encoded, and Among the combinations of geometric division mode candidates and motion information candidates that make up the entire geometric division mode set, regarding the combination of the angle and distance from the block center of the division line segment that is predefined as the entire geometric division mode candidate. , the second feature is that the partial set is determined by using only a part of the combination. Further, the present invention is characterized in that it is a method and a program compatible with the device.

According to the first feature, based on the information on the current block to be encoded, only a partial set selected from the entire geometric division mode set predefined in the geometric division is used for the block to be encoded. By applying inter prediction and geometric division to the image and then using it as an evaluation target for encoding, it is possible to suppress the amount of encoding processing while ensuring quality. According to the second feature, among the combinations of geometric division mode candidates and motion information candidates that constitute the entire geometric division mode set, the dividing line segment is defined in advance as the entire geometric division mode candidate. By using only a part of the combination of angle and distance from the block center, it is possible to similarly suppress the amount of encoding processing while ensuring quality.

1 is a diagram showing an image processing system according to an embodiment. FIG. 1 is a diagram illustrating an example of functional blocks of an image encoding device according to an embodiment. FIG. 3 is a diagram illustrating an example of functional blocks of an inter prediction unit according to an embodiment (reference method based on size information). This is an illustrative example of geometric division. FIG. 2 is an explanatory diagram showing that division boundaries in geometric division are given by a combination of angles and distances. FIG. 2 is a flow diagram of a speeding-up method using the size of a coding block according to an embodiment. FIG. 6 is a diagram illustrating an implementation example of thinning out geometric division mode sets. FIG. 6 is a diagram illustrating an implementation example of thinning out geometric division mode sets. It is a figure which shows an example of the functional block of the inter prediction part based on embodiment of the modification 1 with respect to a reference method. FIG. 7 is a flowchart of a speeding-up method according to an embodiment of modification example 1; FIG. 7 is a flowchart of a speeding-up method according to an embodiment of modification 2; FIG. 7 is a flowchart of a speeding-up method according to an embodiment of modification 3; FIG. 7 is a diagram showing an example of vertical direction and horizontal direction as an example of determining the distance of an angular pattern in Modification 3. FIG. 7 is a flowchart of a speeding-up method according to an embodiment of modification 4; FIG. 11 is a diagram illustrating an example of adjacent blocks referred to when deriving motion information as an explanatory example of selection processing in Modification 4. FIG. 13 is a flowchart of a speeding-up method according to an embodiment of modification 5. FIG. 13 is a flowchart of a speeding-up method according to an embodiment of modification 6. FIG. FIG. 7 is a flowchart of a speeding-up method according to an embodiment of Modification Example 7; FIG. 1 is a diagram showing the hardware configuration of a general computer.

Various embodiments of the present invention will be described below with reference to the drawings. Note that the components in the following embodiments can be replaced with existing components as appropriate, and various variations including combinations with other existing components are possible. Therefore, the content of the invention described in the claims is not limited to the following description of the embodiments.

(First embodiment: encoding processing amount reduction method of geometric division using encoding block size)
An image processing system 10 according to a first embodiment of the present invention will be described below with reference to FIGS. 1 to 8. FIG. 1 is a diagram showing an image processing system 10 according to this embodiment.

(Image processing system 10)
As shown in FIG. 1, the image processing system 10 includes an image encoding device 100 and an image decoding device 200. Image encoding device 100 is configured to generate encoded data by encoding an input image signal. Image decoding device 200 is configured to generate an output image signal by decoding encoded data. Here, the encoded data may be transmitted from the image encoding device 100 to the image decoding device 200 via a transmission path. Furthermore, the encoded data may be generated in the image encoding device 100 and stored in a storage medium, and then provided to the image decoding device 200 by reading this encoded data from the storage medium.

(Image encoding device 100)
The image encoding device 100 according to this embodiment will be described below with reference to FIG. 2. FIG. 2 is a diagram showing an example of functional blocks of the image encoding device 100 according to the present embodiment. The image encoding device 100 includes an inter prediction unit 111, an intra prediction unit 112, a subtracter 121, an adder 122, a transformation/quantization unit 131, an inverse transformation/inverse quantization unit 132, and an encoding unit. 140, an in-loop filter processing section 150, a frame buffer 160, and a block division section 170.

The inter prediction unit 111 is configured to generate predicted pixels corresponding to the encoding target signal by inter prediction (interframe prediction). Specifically, the inter prediction unit 111 uses a frame to be encoded (hereinafter referred to as a target frame) and a reference frame stored in the frame buffer 160 (an already encoded frame (which may be a part of the frame)) to an inverse transform unit. - A reference block included in the reference frame is identified by comparison with a reference frame (as a frame reconstructed by processing after the dequantization unit 132), and a corresponding motion vector (MV) is determined. Furthermore, the inter prediction unit 111 generates predicted pixels for each block to be encoded (hereinafter referred to as a target block) based on the reference block and the motion vector. Inter prediction section 111 outputs predicted pixels to subtracter 121 and adder 122.

The intra prediction unit 112 is configured to generate predicted pixels corresponding to the encoding target signal by intra prediction (intraframe prediction). Specifically, the intra prediction unit 112 identifies a reference block included in the target frame, and generates predicted pixels for each target block based on the reference block. Intra prediction unit 112 outputs predicted pixels to subtracter 121 and adder 122. The reference block is, for example, a block adjacent to the target block.

The subtracter 121 subtracts predicted pixels (predicted pixels predicted by the inter prediction unit 111 or intra prediction unit 112) from the input image signal to generate a prediction residual signal, and converts and quantizes the prediction residual signal. 131.

The adder 122 adds the predicted pixel (the predicted pixel predicted by the inter prediction unit 111 or the intra prediction unit 112) to the prediction residual signal output from the inverse transform/inverse quantization unit 132 to obtain a pre-filtered decoded signal. It is configured to generate a pre-filtered decoded signal to the intra prediction unit 112 and the in-loop filter processing unit 150.

The conversion/quantization unit 131 performs conversion processing and quantization processing on the prediction residual signal output from the subtracter 121 to generate a quantized level value, and inversely transforms and dequantizes the quantized level value. 132 and an encoding unit 140. The conversion process is a process of converting the prediction residual signal into a frequency component signal. In such transformation processing, a basis pattern (transformation matrix) corresponding to a discrete cosine transform (DCT) may be used, and a basis pattern (transformation matrix) corresponding to a discrete sine transform (DST) may be used. may be used. There is no need to perform conversion processing.

The inverse transformation/inverse quantization unit 132 is configured to perform inverse quantization processing and inverse transformation processing on the quantized level value to generate a prediction residual signal, and output it to the addition unit 122.

The encoding unit 140 is configured to encode the quantized level value output from the conversion/quantization unit 131 and output encoded data. The encoding is entropy encoding that assigns codewords of different lengths based on the probability of occurrence of post-quantized level values. Furthermore, the encoding unit 140 is configured to encode control data used in decoding processing in the image decoding device 200 and output encoded data. Here, the control data is additional information for the image decoding device 200 to specify the block division mode, prediction mode, motion information, etc. Although these are omitted from the drawing from the viewpoint of simplifying the drawing, they are output from each functional block of the encoding device 100. For example, the block division mode is output from the block division section 170, the intra prediction mode is output from the intra prediction section 112, and the inter prediction mode and motion information are output from the inter prediction section 111. A control unit (not shown) specifies single control data from a plurality of control data candidates output from each functional block. As a specific method, control data that minimizes rate distortion cost may be selected.

The in-loop filter processing unit 150 performs filter processing on the unfiltered decoded signal, and filters the filtered decoded signal (reconstructed image signal or reference frame, where the reference frame is a reconstructed image signal that is processed by the inter prediction unit 111). (referring to the one used for predictive encoding of the frame) and output it to the frame buffer 160. Filter processing includes deblocking filter processing that reduces distortion occurring at the boundaries of encoding blocks, prediction blocks, or transform blocks, and filter coefficients and filter selection information transmitted from the image encoding device 100, as well as local processing of image patterns. This is an adaptive loop filtering process that switches filters based on characteristics.

Frame buffer 160 is configured to accumulate the filtered decoded signal and output it to inter prediction section 111 as a reference frame.

The block division unit 170 is configured to divide the target frame into blocks each consisting of squares or rectangles that do not overlap with each other, and output the encoding target signal to the inter predictor 111 and the intra predictor 112 for each block. Here, the block division method may be different for the inter prediction unit 111 and the intra prediction unit 112, or may be different for the luminance signal and the color difference signal. Blocks can be divided into quarters, where the vertical and horizontal lengths are halved, two divisions, where the vertical or horizontal lengths are halved, and three divisions, where the vertical or horizontal lengths are 1:2:1. You may also use

(Geometric division mode candidate selection method using the size of the target coding block (hereinafter referred to as the standard method))
(Inter prediction unit 111)
The inter prediction unit 111 according to this embodiment will be explained using FIGS. 3 to 8. FIG. 3 is an example of functional blocks of the inter prediction unit 111 according to this embodiment. The inter prediction unit 111 in FIG. 3 is configured to include a merging unit 1111, a coded block information extraction unit 1112, a geometric division mode candidate selection unit 1113, and a predicted pixel generation unit 1114.

Note that the inter prediction unit 111 in FIG. 3 applies inter prediction to the encoded block, and does not apply inter prediction to the entire encoded block, but applies geometric division to each divided portion. This is a configuration that performs processing when applying inter prediction to. (As mentioned above, it is up to the image encoding device 100 to determine whether or not to apply inter prediction to the encoded block, or whether to apply the geometric division method after applying inter prediction. (The provided control unit (not shown) can perform this separately using any method.)

The merging unit 1111 derives a plurality of pieces of motion information of the target block from the motion information of adjacent positions (adjacent blocks) in space and time with respect to the target coded block, and outputs the motion information.

By using a merge mode technique, the merging unit 1111 extracts a plurality of pieces of motion information as motion information candidates to be applied to the target coded block from the motion information already determined for the coded adjacent blocks. It can be derived and output to the encoded block information extraction unit 1112. (In addition, each piece of motion information that has already been determined for this encoded adjacent block includes a set of motion vectors and an index that specifies a reference frame (hereinafter referred to as reference index) as one piece of motion information. , two sets of motion vectors, and a reference index.Specifically, single prediction using one reference frame is applied to a coding block that includes spatially and temporally adjacent regions. In this case, one set of motion vector and reference index is included, and when bi-prediction using two reference frames is applied or two sets of motion vector and reference index are applied, two sets of motion vector and reference index are included. When geometric partitioning is applied to areas that are adjacent in terms of space and time, only one set of motion vectors and reference indexes are used for predictive coding of the adjacent area based on block partitioning by geometric partitioning. It is also good to obtain.)

Encoding block information extraction section 1112 extracts the size of the target encoding block, and outputs the encoding block information to geometric division mode candidate selection section 1113. Here, the encoding block information includes, for example, (1) the length of the short side of the target encoding block, (2) the length of the long side, or (3) all or any number of pixels of the target encoding block. It can be given as a portion (one or two).

The geometric division mode candidate selection unit 1113 selects a plurality of geometric division mode candidates from the geometric division mode set based on the encoded block information, and selects the geometric division mode candidates from the geometric division mode set. Output. FIG. 4 is an illustrative example of geometric division. In this embodiment, the dividing boundary is expressed as a combination of the angle pattern and distance pattern shown in FIG. 5 as an example.

In FIG. 4, as an example of geometric division (one of multiple division candidates) applied to the target encoded block BL in the frame F to be encoded, the target encoded block BL is divided into a straight line division boundary DB It is shown that it is divided into a first part PA and a second part PB. Note that the first motion vector MVA for the first portion PA and the second motion vector MVB for the second portion PB are given by the merging unit 1111 in the predicted pixel generation unit 1114 at the subsequent stage to optimize the cost. The target is searched from among the motion information of the selected candidates. These motion vectors MVA and MVB constitute a motion vector MV=(MVA, MVB) for each portion PA and PB that are geometrically divided by the division boundary DB. Here, the motion vector MVA points to the block BLA in the reference frame FA as the reference destination of the first part PA, and the motion vector MVB points to the block BLB in the reference frame FB as the reference destination of the second part PB. .

In Fig. 5, the geometric partitioning mode set, which is (all) candidates for the partitioning boundary whose partitioning boundary DB is shown in one case in Fig. 4, is a combination of angle parameter φ and distance (offset) parameter ρ (φ, ρ ) is shown to be given by

The angle parameter φ gives the angle that the straight line of the division boundary DB makes with the block, and the direction from the center of the block to the horizontal right is φ=0°=φ1, and the angle increases in the order shown in the figure. For example, 20 candidates can be prepared such as φ={φ1, φ2, φ3,..., φ20}. For each of these angles φ, 20 different angles φ can be determined so that the possible values of tanφ are, for example, as follows.
tanφ=0, ±1/4, ±1/2, ±1, ±2, ±∞

The distance (offset) parameter ρ gives the displacement of the straight line of the division boundary DB from the block center (displacement in the direction given by the angle parameter φ (parallel movement in the direction of φ + 90°)), and each For each angle φ, four distance parameter candidates can be determined in order of distance parameter ρ={ρ1, ρ2, ρ3, ρ4}, starting from the distance closest to the center. (ρ=ρ1 is the closest to the block center and corresponds to the case where the division boundary DB passes through the block center, and ρ=ρ4 corresponds to the farthest from the block center. As a horizontal movement with respect to the straight line in the direction φ, φ+90 By horizontally moving a straight line from the block center in the direction of ° by the distance parameter ρ, a division boundary DB for each distance parameter ρ is constructed.)

In FIG. 5, for the case of angle parameters φ=φ3, φ9, φ13, φ19=45°, 135°, 225°, 315°, the distance parameter ρ= commonly defined as the distance from the block center in the direction of the angle φ {ρ1, ρ2, ρ3, ρ4} are shown.

As schematically shown in Figures 4 and 5, when applying geometric division, the entire combination to be searched (the entire geometric division mode "set") is the "division boundary" as shown below. Multiplying the number of candidates by the number of "motion information" candidates yields 64*30=1920, which is a huge number. In addition, if there are two sets of motion vectors and an index specifying a reference frame in one motion information, it is determined which of the two sets of motion vectors and an index specifying the reference frame to use using a predetermined method. do. Specifically, it can be determined, for example, by the method described in Non-Patent Document 1.
Division boundary... 64 ways (64 ways excluding the same division boundary and the division boundary that divides the block into four, two, and three in the same way as the 20×4=80 ways in FIG. 5) Become.)
Candidates for motion information...30 types (The merging unit 1111 can acquire six motion information from among the motion information applied during predictive encoding in encoded spatiotemporally adjacent regions.Geometric When dividing, you can select the motion information to be applied to the two regions after dividing the target coded block from these six ways.Since the same motion information cannot be selected to be applied to the two regions, the motion information of the two regions can be selected. The number of combinations is ₆ P ₂ =30.)

The geometric division mode candidate selection unit 1113 of this embodiment narrows down the huge number of candidates in advance using each method described below (narrows down the candidates from all candidates for the geometric division mode), thereby reducing the amount of calculation. can contribute to the reduction of

Returning to the explanation of FIG. 3, the predicted pixel generation unit 1114 generates predicted pixels using the plurality of geometric division mode candidates selected by the geometric division mode candidate selection unit 1113. Here, the predicted pixels may be generated by a weighted average of blocks of each of the two reference frames. The weight may be determined depending on the division boundary.

Note that the predicted pixel generation unit 1114 generates predicted pixels for each of the candidates reduced by the geometric division mode candidate selection unit 1113, and outputs the generated pixels from the inter prediction unit 111. As described above, which geometric division mode is actually used for encoding from among the reduced candidates of geometric division modes is determined by a control unit (not shown) provided in the image encoding device 100. Accordingly, the rate-distortion cost can be determined as the one that minimizes it.

Selection of geometric division mode candidates and prediction signals performed by the encoded block information extraction unit 1112, the geometric division mode candidate selection unit 1113, and the predicted pixel generation unit 1114 of the inter prediction unit 111 of this embodiment The generation flow will be explained using FIG. 6.

In step S1, it is determined in the settings of the image encoding device 100 whether geometric division is effective for the target encoding block. If it is valid, proceed to step S2; if it is invalid, this process ends. (If it is invalid, the target encoding block will be encoded using a predictive encoding method other than the applied geometric division.)

In step S2, the size of the target encoding block is extracted. As described above, all or part of the length of the short side, the length of the long side, and the number of pixels may be used as the size of the target encoding block.

In step S3, a plurality of geometric division mode candidates are selected as candidates narrowed down from the entire geometric division mode set based on the size of the target encoding block.

In step S4, predicted pixels are generated using the geometric division mode candidates, and the flow of FIG. 6 ends.

Details of the selection process in step S3 will be described below.

In this selection process, a predetermined threshold value may be used for the short side length or long side length or area (the number of pixels may be used) of the target encoding block. 1 to 5 can be used.

<Method 1...Reducing the number of candidates when the long side is determined to be short>
For example, geometric division mode candidates may be selected such that the number of candidates when the size of the long side of the target encoding block is less than or equal to a threshold is smaller than the number of candidates when the size of the long side of the target encoding block is larger than the threshold. . Specifically, if it is less than the threshold, multiple geometric division mode candidates are selected from the geometric division mode set, and if it is greater than the threshold, all the geometric division mode sets are selected as geometric division mode candidates. Good too.

I will explain the reason for this. When the target encoding block is small, even if there are few division boundary patterns, the subject, foreground, and background boundaries in the block can be adapted to the limited division pattern, and the encoding performance is maintained while the encoding performance is maintained. It is expected that the amount of processing will be reduced. The threshold value may be set to an arbitrary value, and may be set to 16 pixels with respect to the length of the long side.

<Method 2...Reducing the number of candidates when the short side is determined to be long>
For example, geometric division mode candidates may be selected such that the number of candidates when the size of the short side of the target coded block is equal to or larger than a threshold is smaller than the number of candidates when the size is smaller than the threshold. . Specifically, if it is greater than or equal to the threshold, multiple geometric division mode candidates are selected from the geometric division mode set, and if it is smaller than the threshold, all the geometric division mode sets are selected as geometric division mode candidates. Good too.

The reason for this will be explained. When the target encoding block is large, the boundaries between the subject and the foreground/background within the block are more likely to be composed of a plurality of straight lines than when the block is small. Furthermore, even if the boundaries within the block are composed of one type of straight line, if the block size is large, it is difficult to accurately fit the boundaries within the block at each division boundary in the geometric division mode. Therefore, even if a certain number of similar patterns are reduced from the angle and distance patterns constituting the geometric division mode, the prediction accuracy considering the boundaries within the block due to geometric division does not drop significantly; This can be expected to have the effect of significantly reducing the amount of encoding processing required to select the logical division mode. Note that the threshold value may be set to an arbitrary value, and may be set to 32 pixels with respect to the length of the short side.

<Method 3...Reducing the number of candidates when the area or number of pixels is determined to be small>
Furthermore, in order to select a geometric division mode candidate, the area of the target coding block may be subjected to threshold processing instead of the above-described size of the long side or the short side. If the number is less than or equal to the threshold, geometric division mode candidates may be selected so that the number of candidates is smaller than the number of candidates when the number is greater than the threshold. For example, the number of pixels of the target encoding block may be used as the area of the target encoding block. Further, the threshold value may be set to an arbitrary value, and may be set to 1024 pixels.

<Method 4...Reducing the number of candidates when judged to be long and thin>
In this selection process, a predetermined threshold value may be used for the ratio (aspect ratio) between the length of the short side and the length of the long side of the target encoding block. For example, set geometric division mode candidates so that the number of candidates when the long side length divided by the short side length is greater than or equal to the threshold is smaller than the number of candidates when the result is smaller than the threshold. You may choose.

The reason for this will be explained. When the target encoding block is long and narrow, if the division boundary is configured with the same combination of angle patterns and distance patterns as in the case of a square block, a plurality of similar division boundary patterns will exist. (For example, if it is elongated in the horizontal direction, the angles φ={φ5, φ6, φ7, φ15, φ16, φ17} (near angle φ of tanφ=±∞) in Figure 5 will result in almost the same dividing boundary pattern. ) By reducing similar patterns, it is expected that the amount of encoding processing can be reduced while maintaining encoding performance. Note that the threshold value may be set to an arbitrary value, and may be set to 2 for the result of dividing the length of the long side by the length of the short side.

<Method 5... Thinning out for geometric division mode set>
In this selection process, a geometric division mode of a predetermined pattern (a pattern thinned out from a state in which all patterns are covered) may be selected for the angle pattern and the distance pattern from the geometric division mode set. .

<Implementation example 1 of method 5>
For example, as shown in Figure 7, 8 angles {0°, 45°, 90°, 135°, 180°, 225°, 270°, 315°} are thinned out from a total of 20 angles φ. and/or thinned out in half from all four ways of distance ρ, with only two ways: the second closest dividing boundary and the fourth closest dividing boundary {ρ2,ρ4} from the center. You may also use

In addition, in Figure 7, based on the example in Figure 5, the angle φ is thinned out in 8 ways, and the distance ρ pattern is thinned out in 2 ways {ρ2, ρ4} for angles φ=45° and 90°. Examples are shown.

I will explain the reason for this. Generally, when shooting video, the camera is often held horizontally or vertically to the ground. Therefore, many boundaries of objects such as artificial objects in the video are considered to exist horizontally or vertically to the target block. Here, it is conceivable to make the angle pattern of the division boundary only horizontal and vertical, but it is conceivable that it will not be able to match boundaries that are not horizontal or vertical, resulting in a decrease in prediction accuracy. Therefore, the angle patterns used are horizontal, vertical, and intermediate angles of 45°, 135°, 225°, and 315°. In addition, the geometric division mode in which the angular pattern is horizontal or vertical and uses a distance pattern passing through the center is the same process as dividing into two by the block division unit 170 and then performing predictive encoding on each block. Not present in geometric split mode set to avoid duplication of processing. (That is, as mentioned above, the geometric partitioning mode set is defined to avoid overlap.) Therefore, by setting the distance pattern to the second closest and fourth from the center, the partitioning boundary can be spatially can be reduced evenly. This can be expected to reduce the amount of encoding processing while maintaining encoding performance.

<Implementation example 2 of method 5>
For example, as shown in Figure 8, the angles are 45°, 135°, 225°, 315°, and the distances are between the division boundary passing through the center and the third nearest division boundary, and the angles are 0°, 90°, 180°. 270 degrees, and the second closest dividing boundary and fourth closest dividing boundary from the center may be used. In other words, the following set A (4*2=8 ways) or set B (4*2=8 ways) is 8+8=16 as candidates (φ, ρ) to be narrowed down from a total of 64 geometric division mode sets. It is also possible to use the street.
(φ,ρ)∈A∪B
A={45°,135°,225°,315°}×{ρ1,ρ3}
B={0°,90°,180°,270°}×{ρ2,ρ4}

In addition, in the example of FIG. 8, sets A and B are shown for the angle φ based on the example of FIG. and candidate ρ={ρ2,ρ4} when the angle φ=90° of set B.

The reason for this will be explained. The reason for selecting the angle pattern is as described above. Also, as the distance pattern, you can select one that passes through the center or the one that is third closest to the center only when the distance pattern is 45°, 135°, 225°, or 315°. Here, for combinations such as (45°, 225°) and (135°, 315°) where the angles differ by 180°, the distance pattern passing through the center represents the same division boundary, so only one division boundary is geometrically Present in split mode set. (In other words, as mentioned above, the geometric partitioning mode set is defined to avoid duplication.) Therefore, the selection method shown in FIG. Since the number of elements of geometric division candidates can be reduced compared to the selection method, it is expected that the amount of encoding processing can be further reduced while maintaining encoding performance.

Note that method 5 cannot be used alone, but can be used in combination with any of methods 1 to 4. The case where Method 5 is used alone will be described as Modification 1 below.

(Modification 1: Geometric division mode candidate selection method in smaller configuration)
Hereinafter, with reference to FIGS. 9 and 10, differences between this modified example and the reference method described above will be explained. The difference is that this modification selects geometric division mode candidates without using the size of the target encoding block.

FIG. 9 is a diagram showing the configuration of the inter prediction unit 111 according to the first modification. Differences between FIG. 9 and the configuration shown in FIG. 3 will be explained below. The inter prediction unit 111 in FIG. 9 is configured without the coded block information extraction unit 1112 in FIG. 3 (omitted).

Due to this omission, the geometric division mode candidate selection unit 1113 selects multiple geometric division modes from the geometric division mode set without using the target encoded block information as in the case of the standard method. The geometric division mode candidates are selected and the geometric division mode candidates are output.

FIG. 10 shows the flow of selecting geometric division mode candidates and generating a prediction signal, which are executed by the geometric division mode candidate selection unit 1113 and the predicted pixel generation unit 1114 in this modification. Differences between the flow shown in FIG. 10 and FIG. 6 will be explained below.

The flow in FIG. 10 does not include step S2 in FIG. 5, and the contents of the process in step S13 in FIG. 10, which corresponds to step S3 in FIG. 5, are partially different. Furthermore, steps S1 and S4 in FIG. 10 are the same as steps S1 and S4 in FIG. 5, as indicated by the same reference numerals.

In step S13, a plurality of geometric division mode candidates are selected from the geometric division mode set, regardless of the target encoding block. The selection of geometric division mode candidates is performed in advance with respect to angle patterns and distance patterns, in the same way as explained in the geometric division mode candidate selection method (standard method) using the size of the target coding block. A geometric division mode of a defined pattern may be selected.

That is, in step S13, for example, the method described as method 5 "thinning for geometric division mode set" in the standard method can be applied without using information (size information, etc.) of the target encoded block.

The effects obtained from the above differences are as follows. Even for a target coded block for which a small number of geometric partition mode candidates were not selected in the standard method, the geometric partition mode to be applied to the target coded block is determined from a small number of geometric partition mode candidates. Therefore, further reduction in the amount of encoding processing can be expected.

(Explanation of variations 2 to 7)
Below, as modified examples 2 to 7, six geometric division mode candidate selection methods will be described in which the coded block information to be extracted is changed from the standard method. Although the functional configuration of each method is the same as that of the standard method, the processing contents of the encoded block information extraction unit 1112 are different. Furthermore, the generation flow differs in the processing contents corresponding to step S2 and step S3 in FIG.

(Modification 2: Geometric division mode candidate selection method using distance between target frame and reference frame)
The differences between this modification and the basic selection method will be described below. The encoded information block extraction unit 1112 of this modification extracts the output order (POC: Picture Order Count) of the target frame and outputs it as encoded block information. As is known, POC is the order in which the target frames are played back, and since the frames immediately after reconstruction are not arranged in chronological order, it corresponds to information for rearranging them, and the existing frames are used during encoding. It can be determined automatically using a method.

Next, with reference to FIG. 11, differences in the generation flow of this modified example of the method will be explained. Note that steps S1 and S4 in FIG. 11 have the same processing contents as steps S1 and S4 in FIG. 6, as indicated by the same reference numerals.

In step S22, the POC of the target frame is extracted. In step S23, a plurality of geometric division mode candidates are selected from the geometric division mode set based on the POC of the target frame.

In this selection process, the POC of the target frame and the POC of the reference frame included in the motion information of the geometric division mode set may be subjected to threshold processing. For example, calculate the absolute value of the difference between the POC of the target frame and the POC of the reference frame in one geometric division mode, and if the calculation result is less than the threshold, change the geometric division mode to Select it as one element of the geometric division mode candidates (i.e., treat it not to be excluded from the candidates), and if it is larger than the threshold, do not include it in the geometric division mode candidates (i.e., exclude it from the candidates). good.

In the example of FIG. 4, for the encoding target block BL and the division boundary DB (this division boundary DB and the corresponding motion vector are any one of the entire geometric division mode set), Extract the POC (referred to as POC[F]) of the target frame F that includes this encoding target block BL, and extract the POC (respectively POC[FA],POC[FB]) are extracted and the following is done using the threshold TH.
If |POC(FA)-POC(F)|>TH, the geometric division mode that refers to FA is not used (the same applies to FB).

The reason for this will be explained. If the absolute value of the difference in POC between two frames is large, this corresponds to a case where the two frames are far apart in time. Therefore, in geometric segmentation that predicts the object boundary, the prediction of the animal side is likely to be incorrect, and this is selected as a candidate for geometric segmentation. This is because it is better not to do so.

(Modification 3: Geometric division mode candidate selection method for target coding block using geometric division mode information of neighboring blocks)
The differences between this modified example and the standard method will be described below. The encoded information block extracting unit 1112 extracts geometric division mode information applied to neighboring blocks of the target encoded block and outputs it as encoded block information.

Next, the difference in generation flow will be explained with reference to FIG. 12. Note that Step S1 and Step S4 in the flow of FIG. 12 have the same processing contents as Step S1 and Step S4 in FIG. 6, as indicated by the same reference numerals.

In step S32, geometric division mode information applied to neighboring blocks is extracted. The neighboring blocks may be selected from a plurality of adjacent encoded blocks to which the geometric division mode is applied according to a predetermined rule. For example, a block adjacent to the left and a block above the target encoding block may be selected. If the corresponding block does not exist, information indicating that the corresponding block does not exist is set as geometric division mode information applied to neighboring blocks.

In step S33, a plurality of geometric division mode candidates are selected from the geometric division mode set by one of the following methods 31 to 33, based on the geometric division mode information applied to the neighboring blocks. (Note that method 33 corresponds to skip processing when modification 3 cannot be applied.)

<Method 31>
In this selection process, a geometric division mode in which it is determined that the division boundary of the neighboring block and the division boundary of the target encoded block are continuous may be selected as a geometric division mode candidate. The reason for this will be explained. Generally, most of the boundaries between the subject and the foreground/background that exist in the target frame are continuous. Therefore, by selecting a geometric division mode in which the division boundaries of neighboring blocks and the division boundaries of the target encoding block are continuous as a geometric division mode candidate, it is possible to reduce the amount of encoding processing while maintaining encoding performance. You can expect that.

Examples EX1 and EX2 in FIG. 13 show examples in which division boundaries are determined to be continuous. In example EX1, as the division boundary DB of the target encoding block BL, the division boundaries DBU and DBL applied to two neighboring blocks, the upper adjacent block BLU and the left adjacent block BLL, respectively, and their boundary locations (perfectly) match. This is an example of selecting those that are determined to be continuous by doing this. Similarly, example EX2 is determined as a division boundary DB, and the boundary location is determined to match the adjacent division boundaries DBU and DBL (not an exact match). In this example, the distance between the boundary points is determined to be close based on the threshold value determination, and those determined to be continuous are selected.

<Method 32>
Alternatively, geometric division mode candidates for the target encoding block may be selected with reference to the angle pattern φ of the division boundary (φ, ρ) of neighboring blocks.

Here, an example will be described in which geometric division mode candidates for the target encoding block are selected based on the angle pattern φ being close to each other. Note that the fact that the angle patterns φ are close may be determined based on the fact that |tanφ| is close. In other words, if the difference in the value of φ is ±180° and the directions are opposite, the angle patterns φ are treated as the same, and the closer the difference in the value of φ is ±90°, the larger the difference in the angle patterns. In other words, the closer the difference in the value of φ is to ±0° or ±180°, the smaller the difference in the angular patterns. (In other words, if two angles φa, φb are 180° and opposite directions, they are treated as the same angle, then the absolute value of the difference between the two angles φa, φb can be obtained within the range of 0° to 90°. Therefore, for example, if the absolute value difference is less than the threshold of 45 degrees, it can be treated as a close angle pattern, otherwise it can be treated as a far angle pattern.Alternatively, if the difference is 180 degrees, it can be automatically handled. As a method for determining whether the angles φa and φb are in the same direction, it is also possible to determine the distance of the angles φa and φb based on the magnitude of |cos(φa-φb)|.)

Alternatively, it may be determined whether the angle patterns φ are close or not depending on which preset range the angle patterns φ belong to, and whether they belong to the same range. For example, if the angular pattern of the neighboring block adjacent to the top is vertical, a geometric division mode that is combined from the vertical angular pattern may be selected as a geometric division mode candidate. Furthermore, if the angle pattern of the neighboring block adjacent to the left is horizontal, a geometric division mode that is combined from the horizontal angle pattern may be selected as a geometric division mode candidate.

Note that here, when the angle pattern is in the vertical direction, it refers to an angle pattern within the range of 45° to 135° and 225° to 315°, as shown in example EX3 in Fig. 13, and when it is in the horizontal direction, refers to an angular pattern within the ranges of 0° to 45°, 135° to 225°, and 315° to 360°, as shown in example EX4 in FIG. Although FIG. 13 shows an example of determining angles that are close in the vertical direction (vertical direction) or horizontal direction (horizontal direction), it is also possible to determine close angles in the same way for diagonal angles that are deviated from the vertical or horizontal direction. Can be done.

<Method 33>
In addition, if there is no encoded block adjacent to the target encoding block to which the geometric division mode has been applied, the geometric division mode set (the entire set) is selected as is as a geometric division mode candidate. You may.

(Modification 4: Geometric division mode candidate selection method using the positions of adjacent blocks referred to when deriving motion information)
The differences between this modified example and the standard method will be described below. The encoded information block extraction unit 1112 extracts the position of the referenced adjacent block when the merging unit 1111 derives the motion information of the geometric division mode set, and outputs it as encoded block information.

Next, the difference in generation flow will be explained with reference to FIG. Note that Step S1 and Step S4 in the flow of FIG. 14 have the same processing contents as Step S1 and Step S4 in FIG. 6, as indicated by the same reference numerals.

In step S42, the position of the adjacent block referenced when the merging unit 1111 derives the motion information of the geometric division mode set is extracted. In step S43, a plurality of geometric division mode candidates are selected from the geometric division mode set based on the positions of adjacent blocks referred to when deriving motion information.

Next, with reference to FIG. 15, an example of which geometric division mode from the geometric division mode set is selected as a geometric division mode candidate in this selection process will be described. First, as shown in FIG. 15, each geometric division mode of the geometric division mode set (in FIG. 15, one geometric division mode that gives the division boundary DB as the same example as in FIG. 4 described above is illustrated). ), consider the divided target coded block when the geometric division mode is applied.

At this time, the motion information MVA (see Figure 4) applied to the divided left area PA refers to one of the blocks adjacent to the area PA to be applied, such as three adjacent blocks A0, A1, and A2. When the motion information MVB (see Fig. 4) that is derived in the same way and applied in the same way in the other area PB refers to one of the adjacent blocks, such as two adjacent blocks B0 and B1. , this geometric division mode may be selected as a geometric division mode candidate. (In the example of FIG. 15, only adjacent blocks in a spatial range whose time coincides with the target coded block BL are considered, but similarly, adjacent blocks in a temporal range can also be considered. (Also, adjacent blocks in both time and space may be considered.)

Here, as described above, the merging unit 1111 collects, for example, six pieces of motion information actually used for predictive encoding in spatially and temporally adjacent areas, and collects motion information used for predictive encoding of the target encoding block. output as a candidate (element of the split mode set). For example, for the encoded block BL, it is checked whether there is motion information applied to a block including the adjacent area A0 as an example, and if so, that motion information is used as a candidate, and then the adjacent area A1 is checked... The merging unit 1111 performs processing to check adjacent regions until the number of pieces of motion information reaches six. Therefore, the motion information included in the divided mode set may include information that refers to A0, information that refers to B0, and the like.

Therefore, the geometric division mode in which the motion information applied to PA refers to any of A0, A1, or A2, and the motion information applied to PB refers to any of B1 or B0 is geometric. Modified example 4 can use a method in which a logical division mode candidate may be selected. (In other words, in Modification 4, regarding motion information, as mentioned above, there are ₆ P ₂ = 30 ways in which there are two parts PA, PB (two parts PA, PB divided by any one division boundary DB). Where possible, for partial PA in this division boundary DB, exclude adjacent areas B1, B0, etc. of another portion PB from candidates, and for partial PB, exclude adjacent areas A0, A1, A12, etc. of another portion PA from candidates. By doing this, it is possible to narrow down the motion information candidates to fewer than 30 motion information candidates for each of the 64 division boundary DBs.)

I will explain the reason for this. Consider a case where the target encoded block BL is divided by the line segment of the division boundary DB as shown in FIG. 15. At this time, it can be assumed that the divided areas PA and PB have different motions MVA and MVB. Any one of the motion information MV[A0], MV[A1], MV[A2] derived from the adjacent blocks A0, A1, A2 is the target encoding adjacent to the adjacent block A0, A1, A2. Although it is often possible to predictively encode the area PA on the left side of the block BL with high accuracy, it is considered that it is rare that the area PB on the right side can be predictively encoded with high accuracy.

Therefore, by using the positions of the adjacent blocks referenced as described above, it is possible to select a combination of geometric division modes that can perform predictive coding with high accuracy as geometric division mode candidates, and it is possible to perform coding while maintaining coding efficiency. It can be expected that the amount of processing can be reduced.

In addition, in the example shown in FIG. 15 above, the selection condition for the geometric division mode is that both areas PA and PB refer to any one of the adjacent blocks, but only one of the areas ( Only one of the two divided areas PA, PB) may refer to any one of the adjacent blocks.

(Modification 5: Geometric division mode candidate selection method using prediction direction of intra prediction)
The differences between this modified example and the standard method will be described below. The encoded information block extraction unit 1112 extracts the reference direction for directional prediction of the target encoded block by the intra prediction unit 112, and outputs it as encoded block information. (Note that the target encoding block is encoded by the inter prediction unit 111 applying the geometric division mode according to the present embodiment, but in modification example 5, as an additional process, the processing of the intra prediction unit 112 ( By applying a process that evaluates SAD, SSD, etc. for each reference direction and determines the reference direction that minimizes them, it is possible to narrow down candidates based on the direction of geometric division.)

Next, the difference in generation flow will be explained with reference to FIG. In addition, step S1 and step S4 in the flow of FIG. 16 have the same processing content as step S1 and step S4 of FIG. 6, as indicated by the same reference numerals.

In step S52, a reference direction for directional prediction in the intra prediction unit 112 of the target coded block is extracted. In step S53, a plurality of geometric division mode candidates are selected from the geometric division mode set based on the extracted reference direction.

In this selection method, the reference direction of the directionality prediction of the target coded block in the intra prediction unit 112 is referred to, and the geometric division mode using the angle pattern φ that is close to the reference direction is selected as the geometric division mode. You may select it as a candidate. When the angles are close, it means, for example, that the difference between the respective angles is less than or equal to a threshold value. At this time, the threshold value may be arbitrarily specified, for example, 10 degrees. At this time, as in the case of the above-mentioned modification 3, the difference in angles may be evaluated by treating the opposite directions that differ by 180 degrees as the same direction (angle).

I will explain the reason for this. The reference direction for directional prediction is often determined along the edge or boundary direction of the target encoding block. Therefore, by selecting a geometric division mode with an angular pattern that is close to the reference direction for directional prediction as a geometric division mode candidate, you can select a geometric division mode that is close to the foreground/background or subject boundary. , it is expected that the amount of encoding processing can be reduced while maintaining encoding efficiency.

(Modification 6: Geometric division mode candidate selection method using geometric division mode information of reference block)
The differences between this modified example and the standard method will be described below. The encoded information block extraction unit 1112 extracts the geometric division mode information applied to the reference block of the target encoded block and outputs it as encoded block information.

Next, the difference in generation flow will be explained with reference to FIG. 17. Note that steps S1 and S4 in the flow of FIG. 17 have the same processing contents as steps S1 and S4 in FIG. 6, as indicated by the same reference numerals.

In step S62, geometric division mode information applied to the reference block is extracted. Here, the reference block is a list of all reference blocks (for example, 6 or less) that are referenced by multiple pieces of motion information (for example, 6 ways) obtained by the merging unit 1111 for the current block to be encoded. be. In step S63, a plurality of geometric division mode candidates are selected from the geometric division mode set based on the geometric division mode information (for example, six or less) applied to the reference block.

In this selection process, the geometric division mode of the reference block (the same geometric division mode) may be selected as the geometric division mode candidate of the target encoded block. (In other words, since the number of geometric division modes applied to the reference block of the target encoded block BL is, for example, six or less, the candidates can be narrowed down from 64 as explained in FIG. 5 to six or less. Note that 30 motion information options remain as the search range for optimal encoding.) Also, the geometric division mode of the reference block and the geometric division mode close to the angle pattern are It may be selected as a division mode candidate. For example, the angle patterns being close means that the absolute value of the difference in angle between the two angle patterns is less than or equal to a threshold value. This threshold value may be any numerical value, for example 10°. At this time, as in Modification 3, if the directions are opposite by 180°, they may be treated as the same direction and the difference in angles may be evaluated.

I will explain the reason for this. Many of the reference blocks have pixel values close to those of the target encoded block, and therefore it is thought that the foreground/background or the boundary between subjects are close to each other. Therefore, by selecting the geometric division mode of the reference block as a geometric division mode candidate, it is expected that the amount of encoding processing can be reduced while maintaining encoding efficiency.

(Modification 7: Geometric division mode candidate selection method using Most Probable Mode (MPM) information)
The differences between this modified example and the standard method will be described below. The encoded information block extraction unit 1112 extracts MPM information of the target encoded block applied by the intra prediction unit 112, and outputs it as encoded block information. (Similar to modification 5, MPM information can be extracted by also applying the processing of the intra prediction unit 112 as additional processing to the target coded block in the geometric division mode. , it is possible to check whether the intra prediction unit 112 is applied for actual encoding in an adjacent block, and if it is applied, MPM information may be constructed based on the mode used in that block. )

Next, the difference in generation flow will be explained with reference to FIG. It should be noted that Step S1 and Step S4 in the flow of FIG. 18 have the same processing contents as Step S1 and Step S4 of FIG. 6, as indicated by the same reference numerals.

In step S72, MPM information (consisting of direction information, as in the case of normal intra prediction) of the target coded block is extracted. In step S73, a plurality of geometric division mode candidates are selected from the geometric division mode set based on the MPM information.

In this selection process, depending on the MPM information, a plurality of geometric division modes or one geometric division mode that has been associated in advance based on the MPM information may be selected as a geometric division mode candidate. (In other words, one or more geometric division modes whose angle φ is determined to be close to one direction indicated by the MPM information may be used as candidates. In this case, as in Modification 3, in the opposite direction that differs by 180 degrees, (In this case, the angle difference may be evaluated as the same angle.) Also, a geometric division mode with high encoding efficiency may be estimated using MPM information.

I will explain the reason for this. The MPM information is information on a plurality of prediction modes estimated by the intra prediction unit 112 to be likely to be finally applied to the target coded block. Therefore, it is considered that information about the boundary of the target encoding block and the direction of the texture is often included. Therefore, it is thought that by using MPM information, it is possible to select and estimate a geometric division mode with high coding efficiency. Therefore, by selecting the selected/estimated geometric division mode as a geometric division mode candidate, it is expected that the amount of encoding processing can be reduced while maintaining encoding efficiency.

Note that the above reference method and each of Modifications 1 to 7 may be combined. For example, the standard method and modification 2 may be combined to select geometric division mode candidates in consideration of the size of the target encoding block and the distance between the target frame and the reference frame.

As described above, according to each embodiment of the present invention, in order to determine a small number of geometric division mode candidates from the entire geometric division mode set, instead of using the entire predefined geometric division mode set as candidates as is, Compared to determining the geometric division mode to be applied to the target encoding block from the entire geometric division mode set, it is expected that the amount of encoding processing can be reduced without deteriorating the encoding performance.

Various supplementary examples, alternative examples, additional examples, etc. will be described below.

(1) As an example of the use of the image encoding device 100 according to the embodiment of the present invention, it can be used for encoding during video transmission to a remote location in a remote video conference, and the video quality is Since this technology can speed up the encoding process while ensuring that the system is secure, it is possible to suppress delays in remote conferences and contribute to the realization of remote conferences with a sense of realism. This makes it possible to achieve smooth remote communication, and it is possible to omit the need for the user to actually go to a remote location for meetings, etc. (movement to a remote location is not necessarily required). . Therefore, this example of use of embodiments of the present invention can reduce carbon dioxide emissions by saving energy resources required for user movement, thereby contributing to the Sustainable Development Goals (SDGs) led by the United Nations. It will be possible to contribute to Goal 13: “Take urgent action to combat climate change and its impacts.”

(2) FIG. 19 is a diagram showing an example of the hardware configuration of a general computer device 60. The image encoding device 100 and the image decoding device 200 of this embodiment can be realized as one or more computer devices 60 having such a configuration. Note that when each device is realized by two or more computer devices 60, information necessary for processing may be sent and received via a network. The computer device 60 includes a dedicated processor 60 that is a processor specialized in video encoding and decoding processing, a CPU (central processing unit) 61 that executes predetermined instructions, and a CPU 61 that executes some or all of the instructions executed by the CPU 61. A GPU (graphics processing unit) 62 as a dedicated processor that is executed in conjunction with the CPU 61, a RAM 63 as a main storage device that provides a work area for the CPU 61 (and the dedicated processor 60 and GPU 62), and a ROM 64 as an auxiliary storage device. , a communication interface 65, a display 66, an input interface 67 that accepts user input via a mouse, keyboard, touch panel, etc., and a bus BS for exchanging data between these.

Each functional unit of the image encoding device 100 and the image decoding device 200 and the encoding method and decoding method executed by each device are executed by reading a predetermined program corresponding to the function of each unit, each step, etc. from the ROM 64. It can be realized by all or part of the dedicated processor 60, CPU 61, and GPU 62. Note that the dedicated processor 60, CPU 61, and GPU 62 are all types of arithmetic devices (processors). Here, when display-related processing is performed, the display 66 also operates in conjunction, and when communication-related processing regarding data transmission and reception is performed, the communication interface 65 further operates in conjunction.

100...Image encoding device, 111...Inter prediction unit, 1111...Merge unit, 1112...Encoded block information extraction unit, 1113...Geometric division mode candidate selection unit, 1114...Predicted pixel generation unit

Claims (17)

In an image encoding device that encodes video by applying block-by-block prediction,
When dealing with a block to be coded to which inter prediction and geometric division are applied as the prediction mode, extract the block to be coded information that is information about the block to be coded, and extract the block information to be coded based on the information of the block to be coded. , in order to encode only a partial set selected from the entire geometric division mode set predefined in the geometric division after applying inter prediction and geometric division to the block to be encoded. An image encoding device characterized in that it is used as an evaluation target. The image encoding apparatus according to claim 1, wherein the encoding target block information includes the size of the encoding target block and is extracted. The length of the short side of the block to be encoded is extracted from the encoding target block information, and when the length of the short side is determined to be long, the number of constituent elements of the partial set is determined to be long. 2. The image encoding apparatus according to claim 1, wherein the number of components of the partial set is set to be smaller than the number of components of the partial set when the partial set is not encoded. The length of the long side of the block to be encoded is extracted including the length of the long side of the block to be encoded, and when the length of the long side is determined to be short, the number of constituent elements of the partial set is determined to be short. 2. The image encoding apparatus according to claim 1, wherein the number of components of the partial set is set to be smaller than the number of components of the partial set when the partial set is not encoded. The number of components of a partial set when the encoding target block information is extracted including the aspect ratio between the long side and the short side of the encoding target block, and the encoding target block is determined to be elongated based on the aspect ratio. The image encoding device according to claim 1, wherein the image encoding device is set to be smaller than the number of constituent elements of the partial set when the partial set is not determined to be elongated. When the encoding target block information includes and extracts the size information of the encoding target block given by the area or number of pixels of the encoding target block, and the encoding target block is determined to be small based on the size information. 2. The image encoding apparatus according to claim 1, wherein the number of components of the partial set is set to be smaller than the number of components of the partial set when the partial set is not determined to be small. Extracting the encoding target block information including information on the output order of a target frame including the encoding target block and the output order of reference blocks of the encoding target block,
Motion information of a first part and/or a second part divided by each of the geometric division modes among the combinations of geometric division mode candidates and motion information candidates that constitute the entire geometric division mode set. The image according to claim 1, wherein the partial set is determined by excluding those that are determined to have a large difference between the output order of the candidate reference blocks and the output order of the encoding target block. Encoding device. The encoding target block information includes the geometric division mode that has already been applied to neighboring blocks that are located in the vicinity of the encoding target block and are defined as blocks that have already been encoded. Extract including information,
The partial set is determined to be continuous with the division boundary of the geometric division mode that has already been applied to the neighboring block, or 2. The image encoding apparatus according to claim 1, wherein the image encoding apparatus determines that the angle is determined to be close to the applied geometric division mode. Extracting coded motion information by inter prediction in an adjacent block temporally and spatially adjacent to the current block to be coded as a motion information candidate to be applied to the coded block;
Among the combinations of geometric division mode candidates and motion information candidates that constitute the entire geometric division mode set, the geometric division mode is selected from all the motion information candidates corresponding to each of the geometric division modes. 2. The partial set is determined by selecting only motion information that has been applied to adjacent blocks that are temporally and spatially adjacent to the first portion and the second portion that are divided by the first portion and the second portion, respectively. Image encoding device. Extracting a reference direction when applying intra prediction to the encoding target block is performed as an additional process,
extracting information on the extracted reference direction in the encoding target block information;
The image encoding apparatus according to claim 1, wherein the partial set is determined as a set of geometric division modes whose angle is determined to be close to the reference direction. Extracting coded motion information by inter prediction in an adjacent block temporally and spatially adjacent to the current block to be coded as a motion information candidate to be applied to the coded block;
Extract the reference block referenced by the motion information,
Among the combinations of geometric division mode candidates and motion information candidates that make up the entire geometric division mode set, geometric division mode candidates are limited to only the geometric division modes that have been applied to the reference block. The image encoding apparatus according to claim 1, wherein the partial set is determined by performing the following steps. Extracting the most probable mode when applying intra prediction to the encoding target block is performed as an additional process,
extracting information on the extracted most probable mode in the encoding target block information;
The image encoding apparatus according to claim 1, wherein the partial set is determined as a set of geometric division modes whose angle is determined to be close to the direction of the most probable mode. Among the combinations of geometric division mode candidates and motion information candidates that constitute the entire geometric division mode set, the angle and distance from the block center of the dividing line segment defined in advance as the entire geometric division mode candidate. 2. The image encoding apparatus according to claim 1, wherein the partial set is determined by using only a part of the combination. In an image encoding device that encodes video by applying block-by-block prediction,
When dealing with a block to be encoded to which inter prediction and geometric division are applied as prediction modes, only a subset selected from the entire geometric division mode set predefined in the geometric division is used. Used to encode after applying inter prediction and geometric division to the block to be encoded,
Among the combinations of geometric division mode candidates and motion information candidates that constitute the entire geometric division mode set, the angle and distance from the block center of the dividing line segment defined in advance as the entire geometric division mode candidate. An image encoding apparatus characterized in that the partial set is determined by using only a part of the combination. In an image encoding method that encodes video by applying block-by-block prediction,
When dealing with a block to be coded to which inter prediction and geometric division are applied as the prediction mode, extract the block to be coded information that is information about the block to be coded, and extract the block information to be coded based on the information of the block to be coded. , in order to encode only a partial set selected from the entire geometric division mode set predefined in the geometric division after applying inter prediction and geometric division to the block to be encoded. An image encoding method characterized in that an evaluation target is used. In an image encoding method that encodes video by applying block-by-block prediction,
When dealing with a block to be encoded to which inter prediction and geometric division are applied as prediction modes, only a subset selected from the entire geometric division mode set predefined in the geometric division is used. Used to encode after applying inter prediction and geometric division to the block to be encoded,
Among the combinations of geometric division mode candidates and motion information candidates that constitute the entire geometric division mode set, the angle and distance from the block center of the dividing line segment defined in advance as the entire geometric division mode candidate. An image encoding method characterized in that the partial set is determined by using only a part of the combination. A program for causing a computer to function as an image encoding device according to any one of claims 1 to 14.

Images