JP2022103284A - Video encoder, video decoder, and their program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a video encoder, a video decoder, and their programs configured to allow for improvement in compression efficiency related to the positive and negative signs of a differential vector in video encoding system using motion compensation prediction.

SOLUTION: A video encoder 1 includes a finite vector code estimation part 13 generating an estimation code of differential vector according to a restriction predetermined for the differential vector related to a prescribed block, a right/wrong flag generation part 14 generating a right/wrong flag indicating authenticity of estimation code, and a context adaptive reversible compression part 15 for reversibly compressing the right/wrong flag. The video decoder 2 includes a finite difference vector code estimation part 27 corresponding to the finite vector code estimation part 13, a context adaptive reversible decoding part 26 for decoding the compressed right/wrong flag, a finite difference vector code determination part 28 for determining the positive negative code of the finite difference vector by using the decoded right/wrong flag and the estimation code, and a finite difference vector restoration unit 22 for restoring the finite difference vector.

SELECTED DRAWING: Figure 1

COPYRIGHT: (C)2022,JPO&INPIT

Description

The present invention relates to a video coding device, a video decoding device, and a program thereof in a video coding method using motion compensation prediction.

Generally, in a video coding method using motion compensation prediction, it is considered meaningless to perform compression coding because the positive and negative codes (+,-) of motion vectors and difference vectors are not biased in appearance frequency, and are compressed. (Lossless compression) is not performed.

For example, in the MPEG-H HEVC / H.265 method, the motion vector is expressed as the sum of the prediction vector (PMV) and the difference vector (MVD) (see, for example, Non-Patent Document 1).

The expression of the difference vector differs depending on whether the absolute value of the vector is 0 or other than 0. This is represented by abs_mvd_greater0_flag, and when abs_mvd_greater0_flag is 0, the difference vector is 0. On the other hand, when abs_mvd_greater0_flag is 1, the difference vector is represented by a combination of the absolute value of the difference vector and the positive / negative sign of the difference vector called mvd_sign_flag. Since the + and-of these positive and negative codes are generally not biased in appearance frequency, mvd_sign_flag does not perform lossless compression, but +/- of the positive and negative codes is used as a 1/0 bit representation, and in bypass mode without lossless compression. It is arithmetically encoded and transmitted in association with the absolute value of the difference vector. That is, in the MPEG-H HEVC / H.265 method, the positive and negative signs of the difference vector are not compressed at all.

ITU-T Recommendation H.265 (2013), High efficiency video coding; ISO / IEC 23008-2: 2013, Information technology High efficiency coding and media delivery in heterogeneous environments Part 2: High efficiency video coding, 2003, 〈https: / /www.iso.org/standard/35424.html>

As described above, in a general video coding method represented by the MPEG-H HEVC / H.265 method or the like, there is no bias in the appearance frequency of the positive and negative codes of the difference vector, so that there is no point in performing compression coding. It is not compressed (lossless compression).

However, depending on the position of the block of interest in the frame, the probability of occurrence of the positive and negative signs (+,-) of the difference vector may also be biased, so compression may be possible by context-adaptive lossless compression. be. Therefore, a technique that enables compression (lossless compression) for the positive and negative codes (+,-) of the difference vector and further improves the compression efficiency related to video coding is desired.

Therefore, an object of the present invention is an image coding device configured to improve the compression efficiency related to the positive and negative codes (+,-) of the difference vector in the image coding method using motion compensation prediction in view of the above-mentioned problems. , A video decoder, and programs thereof.

The video coding device of the present invention is a video coding device in a video coding method using motion compensation prediction, and is a difference vector generation means for generating a difference vector based on motion compensation prediction for a predetermined block in a frame. , The estimated code of the difference vector by estimating the positive and negative signs of the difference vector according to a predetermined rule based on the information about the predetermined block, the absolute value of the difference vector, and the prediction vector related to the difference vector. The difference vector code estimation means for determining the above, the positive and negative codes of the difference vector obtained from the difference vector generation means, and the estimation code of the difference vector obtained from the difference vector code estimation means are compared and both are equal. A correct / incorrect flag generation means that generates a correct / incorrect flag indicating true if the case is true and false if different, and the correct / incorrect flag generated by the correct / incorrect flag generation means are reversible adapted to the context according to the position of the predetermined block. It is provided with a context-adaptive reversible compression means for performing compression processing and a stream generation means for generating a stream for transmitting the correct / incorrect flag subjected to the reversible compression processing in association with the absolute value of the difference vector. It is characterized by.

Further, in the video coding apparatus of the present invention, the difference vector code estimation means calculates positive and negative candidate vectors using the information about the predetermined block, the absolute value of the difference vector, and the prediction vector. Then, depending on the positional relationship between the position of the block corresponding to each candidate vector and the screen edge of the frame, or the coordinates and size of the predetermined block, the positive and negative candidate vectors, and the said. It is characterized in that the positive / negative sign of the difference vector is estimated based on a predetermined rule divided into cases according to the positional relationship of the screen width of the frame.

Further, in the video coding apparatus of the present invention, the frame is divided into a region in the center of the screen and a region within a predetermined range at the edge of the screen, and the predetermined block is contained in at least a region within the predetermined range at the edge of the screen. At that time, it is further provided with a control means for controlling the operation of the correctness flag generation means.

Further, in the video coding apparatus of the present invention, the control means has the difference vector based on which region of the regions within the predetermined range of the screen edge divided into a plurality of regions contains the predetermined block. It is characterized by having a means for controlling to switch the component using the correctness flag for each of the horizontal component and the vertical component of the above.

Further, in the video coding apparatus of the present invention, when the predetermined block is closer to the left and right sides than the upper and lower sides of the screen of the frame, the control means has a horizontal component of the difference vector with respect to the difference vector code estimation means. If the predetermined block is closer to the upper and lower sides than the left and right sides of the screen of the frame, the positive and negative codes of the vertical components of the difference vector are determined first by the difference vector code estimation means. When the predetermined block is at equal distances from the left and right sides and the top and bottom sides of the screen of the frame, the positive and negative codes of the horizontal component and the vertical component of the difference vector are assigned to the difference vector code estimation means in a predetermined order. It is characterized by having a means for controlling to make a decision.

Further, the video decoding device of the present invention inputs and reads a stream transmitted by the video coding device of the present invention, and performs the absolute value of the difference vector and the reversible compression process for a predetermined block in the frame to be decoded. A positive / negative sign of the difference vector according to a predetermined rule based on a stream reading means for extracting the applied correctness flag, information about the predetermined block, an absolute value of the difference vector, and a prediction vector related to the difference vector. By estimating, the difference vector code estimation means for determining the estimated code of the difference vector and the correctness flag subjected to the reversible compression process are reversible adapted to the context according to the position of the predetermined block. A context-adaptive reversible decoding means that generates a decoded correct / incorrect flag by performing a decoding process, and when the decoded correct / incorrect flag indicates true, the estimated code of the difference vector is used as the difference vector of the predetermined block. When a positive / negative sign is used and a false sign is indicated, a difference vector code determining means for determining a code obtained by inverting the estimated code of the difference vector as a positive / negative code of the difference vector, and the extraction to restore the image of the predetermined block. A difference vector restoration means for restoring the difference vector of the predetermined block in the frame to be decoded by using the absolute value of the difference vector and the positive / negative sign of the difference vector determined by the difference vector sign determining means. It is characterized by having.

Further, in the video decoding device of the present invention, the context-adaptive lossless decoding means refers to a block at a predetermined position as a block to be referred to as a context according to the position of the predetermined block in the screen. It is characterized in that the lossless decoding process is performed on the correct / incorrect flag to which the lossless compression process is performed.

Further, in the video decoding device of the present invention, the context-adaptive reversible decoding means is a block referred to as a context according to the position of the predetermined block in the screen, and the predetermined block is a screen rather than the upper and lower sides of the screen. If it is close to the left and right sides, it refers to the block directly above the predetermined block, and if the predetermined block is closer to the upper and lower sides of the screen than the left and right sides of the screen, it refers to the block directly to the left of the predetermined block. When the block of is at an equal distance from the upper and lower sides of the screen than the left and right sides of the screen, the reversible of the block is referred to one or more predetermined blocks of the predetermined block directly to the left, to the upper left, to the upper left, and to the upper right of the screen. It is characterized in that the reversible decoding process is performed on the correct / incorrect flag that has been compressed.

Further, the program of the present invention is configured as a program for operating the computer as the video coding apparatus of the present invention.

Further, the program of the present invention is configured as a program for operating the computer as the video decoding device of the present invention.

According to the present invention, lossless compression is possible with respect to the positive and negative codes (+,-) of the difference vector in the video coding method using motion compensation prediction, and attention is paid to the prediction target located at least at the screen edge of the frame constituting the video. The compression efficiency related to the positive and negative signs (+,-) of the difference vector in the block can be improved.

It is a block diagram which shows the schematic structure of the image coding apparatus of one Embodiment by this invention. It is a block diagram which shows the schematic structure of the image decoding apparatus of one Embodiment by this invention. It is a flowchart which shows the estimation process of the positive / negative code of the difference vector of Example 1 in the image coding apparatus and the image decoding apparatus of one Embodiment by this invention. (A) and (b) are explanatory views of the process relating to the estimation process of the positive / negative code of the difference vector of Example 1 in the image coding apparatus and the image decoding apparatus of one Embodiment by this invention. (A), (b), and (c) correspond to the block of interest, the candidate vector, and the candidate vector related to the estimation processing of the positive and negative signs of the difference vector of Example 1 in the video decoding apparatus of one embodiment according to the present invention, respectively. It is explanatory drawing about the block to do. It is explanatory drawing at the time of a pan operation which concerns on the estimation processing of the positive and negative code of the difference vector of Example 1 in the video coding apparatus and the video decoding apparatus of one Embodiment by this invention. A frame having a region for using a correct / incorrect flag according to the video coding device and the video decoding device according to the present invention and a region where +/- of a positive / negative code is represented by 1/0 as in the conventional technique. It is explanatory drawing. It is a flowchart which shows the determination process of the positive / negative sign of the difference vector of one Example in the video decoding apparatus of one Embodiment by this invention. (A), (b), and (c) are explanatory views of a block referred to as a context related to a context-adaptive lossless compression / decoding process in the video coding device and the video decoding device of one embodiment according to the present invention. .. It is a flowchart which shows the estimation process of the positive / negative code of the difference vector of Example 2 in the image coding apparatus and the image decoding apparatus of one Embodiment by this invention. It is a flowchart which shows the estimation process of the positive / negative code of the difference vector of Example 3 in the image coding apparatus and the image decoding apparatus of one Embodiment by this invention.

(Definition of terms)
First, regarding the motion vector, when the motion vector itself is transmitted as in the MPEG-2 / H.262 method, or when it is transmitted as the sum of the prediction vector and the difference vector as in the MPEG-H HEVC / H.265 method, etc. There is. Therefore, in order to generalize the motion vector used in the video coding device 1 and the video decoding device 2 according to the present invention, the motion vector will be described as satisfying the equation (1).

Motion vector = Prediction vector + Difference vector (1)

The "prediction vector" in the equation (1) means data in which a predetermined number or the like is transmitted without transmitting the vector itself, and a vector is calculated from the number or the like according to a predetermined rule. Further, the "difference vector" in the equation (1) means data for transmitting the vector itself.

Since the present invention is not directly related to the prediction vector of the equation (1) but directly to the positive and negative codes of the difference vector, the image of the embodiment according to the present invention described later with reference to FIGS. 1 and 2. In the coding device 1 and the video decoding device 2, only the main components related to the transmission of the positive and negative codes of the difference vector will be illustrated and described.

In the case of the MPEG-H HEVC / H.265 method, the prediction vector of the formula (1) is the prediction vector (PMV) of the MPEG-H HEVC / H.265 method, and the difference vector of the formula (1) is the MPEG-H HEVC. It corresponds to the difference vector (MVD) of the / H.265 method as it is, and the video coding device 1 and the video decoding device 2 according to the present invention can be applied.

Further, in the case of the video coding method that does not use the prediction vector, the prediction vector of the formula (1) can be treated as 0 and the difference vector of the formula (1) can be treated as the motion vector of the method, without losing the generality. The video coding device 1 and the video decoding device 2 according to the present invention can be applied.

Further, in recent years, in order to calculate the motion vector, a video coding method in which a third vector related to the calculation of the prediction vector is added in addition to the prediction vector and the difference vector has been studied. The prediction vector of (1) can be treated as the prediction vector of the method + the third vector of the method, and the difference vector of the formula (1) can be treated as the difference vector of the method. The video coding device 1 and the video decoding device 2 according to the present invention can be applied.

Hereinafter, the video coding device 1 and the video decoding device 2 according to the embodiment of the present invention will be described with reference to FIGS. 1 and 2.

(Video coding device)
FIG. 1 is a block diagram showing a schematic configuration of a video coding device 1 according to an embodiment of the present invention. The video coding device 1 includes an image coding unit 10, a difference vector code estimation unit 13, a correct / incorrect flag generation unit 14, a context-adaptive lossless compression unit 15, a positive / negative binarization unit 16, a switching unit 17, and a setting control unit 18. And a stream generation unit 19. The image coding unit 10 includes a motion compensation prediction unit 11 and a difference vector generation unit 12.

The image coding unit 10 is a functional unit that inputs a frame image of a video to be encoded, divides it into blocks, and performs intra-prediction and inter-prediction in block units. The present invention relates to compression / decoding of a positive / negative sign of a difference vector based on motion compensation prediction according to inter-prediction. Therefore, in FIG. 1, simply assuming that the image coding unit 10 has a function of generating a difference vector based on the motion compensation prediction for the block of interest, only the motion compensation prediction unit 11 and the difference vector generation unit 12 are illustrated and predicted. Illustrations and explanations of other processes including vector transmission processing and local decoding processing are omitted.

The motion compensation prediction unit 11 performs motion compensation prediction based on motion detection using the block of interest of the input image and the locally decoded reference block, generates a block of the predicted image, and outputs the block to the difference vector generation unit 12. do.

Further, the difference vector generation unit 12 approximates the motion vector generated by the motion detection with a predetermined number such as an index for the motion vector indicating the position of the block of the predicted image as shown in the equation (1). Generate a difference vector that adds a vector to the predicted vector.

Here, the image coding unit 10 outputs the difference vector generated by the difference vector generation unit 12 separately into the absolute value of the difference vector and the positive / negative sign of the difference vector.

The difference vector code estimation unit 13 inputs from the image coding unit 10 the block information of interest (including the position, size, and size of the frame image of the block of interest), the prediction vector, and the absolute value of the difference vector regarding the block of interest. Then, the positive / negative sign of the difference vector is estimated based on a predetermined rule (that is, "estimation processing of the positive / negative sign of the difference vector" described later), and the difference vector estimation code indicating the estimated positive / negative sign is set as a correct / incorrect flag. Output to the generation unit 14.

The correct / incorrect flag generation unit 14 compares the difference vector estimation code obtained from the difference vector code estimation unit 13 with the positive / negative code of the difference vector obtained from the image coding unit 10, and if both are equal, "true". If they are different, a correct / incorrect flag indicating "false" is generated and output to the context-adaptive reversible compression unit 15.

The context-adaptive lossless compression unit 15 inputs the relevant block information of interest from the image coding unit 10, inputs the correct / incorrect flag from the correct / incorrect flag generation unit 14, specifies a context model based on the block information of interest, and determines the context model. By performing a context-adaptive lossless compression coding process on the correct / incorrect flag, a losslessly compressed correct / incorrect flag is generated and output to the switching unit 17. For the context-based lossless compression coding process, for example, in the case of a video coding method according to the MPEG-H HEVC / H.265 method, CABAC (Context-based Adaptive Binary Arithmetic Coding) is used as AVC / H.264. CABAC or CAVLC (Context-based Adaptive Variable Length Coding) can be used when the video coding method conforms to the method. That is, the context-adaptive lossless compression unit 15 performs coding according to the context model and adaptively binarizing according to the surrounding situation (context) at the position of the block of interest.

However, in the video coding apparatus 1 of the present embodiment, as a context model related to lossless compression of the correct / incorrect flag, details will be described later with reference to FIG. 9, but which block around the block of interest is referred to as a context. There are good examples.

The positive / negative binarization unit 16 performs binarization processing in which +/- of the positive / negative sign is represented by a bit of 1/0, and outputs the result to the switching unit 17, as in the conventional technique.

Under the control of the setting control unit 18, the switching unit 17 transmits the positive / negative code of the difference vector of the block of interest by the correct / incorrect flag reversibly compressed by the context-adaptive lossless compression unit 15, or the positive / negative binarization unit 16. It is a functional unit that switches whether to transmit by simply binarizing positive and negative codes and outputs it to the stream generation unit 19.

The setting control unit 18 has an area for designating a context model according to the "estimation process of the positive / negative sign of the difference vector" according to the present invention according to the position of the block of interest in the frame image, and the positive / negative code as in the conventional technique. It has setting information in which a region in which +/- is expressed as 1/0 bit is predetermined, and controls switching in the switching unit 17 based on the setting information. In this example, an example in which the setting control unit 18 outputs the setting information to the stream generation unit 19 will be described, but when the setting information is predetermined between the video coding device 1 and the video decoding device 2, the video code is used. It is not necessary to transmit from the computer 1 to the video decoding device 2.

Note that FIG. 1 shows a configuration example in which the setting control unit 18 controls the switching in the switching unit 17 and switches the output of the context-adaptive lossless compression unit 15 and the positive / negative binarization unit 16. The operation of the code estimation unit 13, the correct / incorrect flag generation unit 14, and the context-adaptive lossless compression unit 15 may be switched between the operation of the positive / negative binarization unit 16.

The stream generation unit 19 determines the correct / incorrect flag after the lossless compression obtained from the switching unit 17 or the simply binarized positive / negative with respect to the absolute value of the difference vector corresponding to the block of interest obtained from the image coding unit 10. Generates a stream for transmission by associating the code and outputs it to the outside. In this example, the setting information obtained from the setting control unit 18 is the header (slice or tile header, picture) in the stream together with the block information of interest (including the position, size, and size of the frame image of the block of interest). And one or more of the headers of the sequence). The stream also includes information necessary for decoding such as a prediction vector. Since any method based on the conventional technique can be used for transmission of the difference vector other than the positive and negative signs, the description thereof will be omitted.

As described above, the video coding apparatus 1 generates a difference vector based on motion compensation prediction for the block of interest, and regarding the positive and negative signs of the difference vector, the block information of interest, the absolute value of the difference vector, and the prediction vector related to the difference vector. The positive and negative signs of the difference vector are estimated according to a predetermined rule based on (that is, "estimation processing of the positive and negative signs of the difference vector" described later). Then, the video coding apparatus 1 sets the positive / negative sign (difference vector estimation code) estimated for the difference vector to be true if it is equal to the positive / negative sign of the difference vector obtained at the time of generating the difference vector, and false if it is different. Is generated, and a context-adaptive reversible compression process is performed for transmission.

As a result, in the case of a block of interest in a frame image in which the occurrence probability is biased, the positive / negative sign (+,-) of the difference vector is expressed by a correct / incorrect flag, and its compression (lossless compression) becomes possible, resulting in a more video code. It is possible to improve the compression efficiency related to the conversion.

(Video decoder)
FIG. 2 is a block diagram showing a schematic configuration of the video decoding apparatus 2 according to the embodiment of the present invention. The video decoding device 2 includes a stream reading unit 20, an image decoding unit 21, a setting control unit 24, a switching unit 25, a context-adaptive reversible decoding unit 26, a difference vector code estimation unit 27, a difference vector code determination unit 28, and positive / negative determination. A unit 29 is provided. The image decoding unit 21 includes a difference vector restoration unit 22 and a motion compensation prediction unit 23.

The stream reading unit 20 reads the stream transmitted from the video coding device 1, outputs a signal relating to the absolute value of the difference vector in the block of interest to the image decoding unit 21 and the difference vector code estimation unit 27, and outputs the signal related to the absolute value of the difference vector to the image decoding unit 21 and the difference vector code estimation unit 27. The correct / incorrect flag after reversible compression or the simply binarized positive / negative code transmitted in association with the absolute value is output to the switching unit 25.

The image decoding unit 21 is a functional unit that performs intra-prediction and inter-prediction in block units to perform decoding. The present invention relates to compression / decoding of a positive / negative sign of a difference vector based on motion compensation prediction according to inter-prediction. Therefore, in FIG. 2, briefly, the image decoding unit 21 has a function of performing motion compensation prediction using the difference vector, the prediction vector, and the decoded reference block restored for the block of interest, and restoring the transmitted image. Only the difference vector restoration unit 22 and the motion compensation prediction unit 23 are shown, and the illustration and description of other processes including the prediction vector decoding process are omitted.

The difference vector restoration unit 22 is a correct / incorrect flag after reversible compression or simply, which is transmitted in association with the absolute value of the difference vector input from the stream reading unit 20 and the absolute value of the difference vector. The difference vector is restored using the binarized positive and negative codes.

Further, the motion compensation prediction unit 23 performs motion compensation prediction using the difference vector restored by the difference vector restoration unit 22, the transmitted prediction vector, and the decoded reference block, and restores the image of the block of interest.

The setting control unit 24 has an area for designating a context model according to the "estimation process of the positive / negative sign of the difference vector" according to the present invention according to the position of the block of interest in the frame image, and the positive / negative code as in the conventional technique. The switching in the switching unit 25 is controlled based on the preset setting information in the area where +/- is represented by 1/0. In this example, the setting information is embedded in a header in the stream (including one or more of a slice or tile header, a picture header, and a sequence header). Therefore, the setting control unit 24 extracts the setting information and controls switching in the switching unit 25 based on the extracted setting information. When the setting information is predetermined between the video coding device 1 and the video decoding device 2, it is not necessary to transmit the setting information from the video coding device 1 to the video decoding device 2, and it is necessary to embed the setting information in the header. Nor.

Under the control of the setting control unit 24, the switching unit 25 switches between the positive / negative sign of the losslessly compressed correct / incorrect flag and the simply binarized difference vector, and outputs them to the context-adaptive lossless decoding unit 26 and the positive / negative determination unit 29, respectively. It is a functional part.

In this example, FIG. 2 shows a configuration example in which the setting control unit 24 controls switching in the switching unit 25 and switches inputs to the context-adaptive reversible decoding unit 16 and the positive / negative determination unit 29, which will be described later, respectively. However, the operation of the context-adaptive reversible decoding unit 26, the difference vector code estimation unit 27, and the difference vector code determination unit 28, which will be described later, may be switched between the operation of the positive / negative determination unit 29, respectively.

The context-adaptive lossless decoding unit 26 inputs the attention block information from the stream reading unit 20 and the losslessly compressed correct / incorrect flag on the video encoding device 1 side via the switching unit 25, and is based on the attention block information. By designating a context model and performing a context-adaptive lossless decoding process on the correct / incorrect flag, the decoded correct / incorrect flag is generated and output to the difference vector coding determination unit 28. For the context-adaptive reversible decoding process, for example, CABAC can be used when the video coding method conforms to the MPEG-H HEVC / H.265 method, while the video coding conforms to the AVC / H.264 method. In the case of the method, CABAC or CAVLC can be used. That is, the context-adaptive reversible decoding unit 26 performs decoding that is adaptively binarized according to the surrounding situation (context) at the position of the block of interest according to the context model. Further, the block information of interest (including the position, size, and size of the frame image of the block of interest) is one or more of the headers (slice or tile header, picture header, and sequence header) in the stream. Is embedded in).

However, in the video decoding device 2 of the present embodiment, as a context model related to lossless compression of the correct / incorrect flag, details will be described later with reference to FIG. 9, but it is suitable for which block around the block of interest is referred to as the context. There is an example.

The difference vector code estimation unit 27 performs the same processing as the difference vector code estimation unit 13 on the video coding device 1 side. The difference vector code estimation unit 27 inputs the block information of interest and the absolute value of the difference vector from the stream reading unit 20, inputs the prediction vector from the image decoding unit 21, and the predetermined rule (that is, the “difference” described later). Based on the vector positive / negative sign estimation process ”), the positive / negative sign of the difference vector is estimated, and the difference vector estimation code indicating the estimated positive / negative sign is output to the difference vector code determination unit 28.

When the correctness flag obtained from the context-adaptive reversible decoding unit 26 indicates "true", the difference vector code determination unit 28 uses the difference vector estimation code obtained from the difference vector code estimation unit 27 as the positive / negative code of the difference vector as it is. It is determined and output to the image decoding unit 21, and when the correctness flag indicates "false", the code obtained by reversing the difference vector estimation code is determined as the positive / negative code of the difference vector and output to the image decoding unit 21.

The positive / negative determination unit 29 corresponds to a bit representation of 0/1 of the positive / negative sign of the simply binarized difference vector obtained via the switching unit 25, and the positive / negative code of the difference vector is the same as in the conventional technique. +/- Is determined and output to the image decoding unit 21.

As a result, the difference vector restoration unit 22 in the image decoding unit 21 has been transmitted in association with the absolute value of the difference vector input from the stream reading unit 20 and the absolute value of the difference vector, after the reversible compression. The difference vector is restored using the correctness flag or the simply binarized positive / negative sign.

As described above, the video decoding device 2 determines the positive / negative sign of the difference vector related to the block of interest read from the stream based on the predetermined rule (that is, “estimation processing of the positive / negative sign of the difference vector” described later). The positive and negative signs of the vector are estimated, and the losslessly compressed positive / false flag read from the stream and subjected to lossless compression is losslessly decoded. Then, the video decoding device 2 uses the estimated positive / negative sign (difference vector estimated code) as the positive / negative sign of the difference vector when the correct / incorrect flag indicates true, and when the positive / negative code indicates false, the estimated positive / negative code (difference). The sign obtained by inverting the vector estimation code) is determined as the positive / negative sign of the difference vector.

As a result, in the case of a block of interest in a frame image in which the occurrence probability is biased, it becomes possible to decode (losslessly compress) the correct / incorrect flag compressed (lossless compression) for the positive and negative codes (+,-) of the difference vector. It is possible to further improve the compression efficiency related to video coding.

(Example 1: Estimating processing of positive / negative sign of difference vector)
FIG. 3 shows the estimation of the positive and negative codes of the difference vector of Example 1 in the video coding device 1 (difference vector code estimation unit 13) and the video decoding device 2 (difference vector code estimation unit 27) of the embodiment according to the present invention. It is a flowchart which shows the process.

First, the difference vector code estimation units 13 and 27 input (read) the absolute value of the difference vector (step S11) for the block of interest, and determine whether or not the absolute value of the difference vector = 0 (step S12). ..

As for the method of reading the absolute value of the difference vector from the bit stream, the same method as the conventional technique may be used. For example, reading the absolute value of the Golomb-coded difference vector from the bitstream, or a flag indicating that the absolute value of the difference vector is larger than 0 as in the HEVC / H.265 method, and a flag indicating that it is larger than 1. , The absolute value of the Golomb-coded difference vector may be read in combination.

When the absolute value of the difference vector = 0 (step S12: Y), the difference vector code estimation units 13 and 27 end the estimation process assuming that there is no positive or negative sign of the difference vector (step S13). When the absolute value of the read difference vector is 0, it is not necessary to determine the positive / negative sign. Therefore, the difference vector = 0 without performing the subsequent processing, and the estimation process is terminated with no positive / negative sign of the difference vector.

On the other hand, when the absolute value of the difference vector ≠ 0 (step S12: N), the difference vector code estimation units 13 and 27 first calculate the positive and negative candidate vectors _VA and V _B for the block of interest. (Step S14).
Candidate vector _VA = Prediction vector + Absolute value of difference vector Candidate vector VA = Prediction vector _- Absolute value of difference vector

The prediction vector can be obtained by a method according to a conventional technique such as HEVC / H.265 method. The prediction vector may be obtained before the absolute value of the difference vector or the correctness flag.

Subsequently, the difference vector code estimation units 13 and ₂₇ calculate the block BK _A and the block BK _B in which the block BKn of interest is moved to the candidate vector _VA and the candidate vector VA, respectively (step S15).

Subsequently, the difference vector code estimation units 13 and 27 determine whether the blocks BK _A and BK _B are both completely in the screen (step S16).

When both blocks BK _A and BK _B are completely in the screen (step S16: Y), the difference vector code estimation units 13 and 27 estimate the positive and negative signs of the difference vector as default values, and the difference vector estimation code = It is determined as a default value (step S17). In this example, "positive" is used as the default value, but "negative" may be used as the default value.

The default value of the difference vector estimation code according to step S17 is the header in the stream (including one or more of the header of the slice or tile, the header of the picture, and the header of the sequence) as the setting information in this example. Embedded in. However, when the default value is predetermined between the video coding device 1 and the video decoding device 2, it is not necessary to transmit from the video coding device 1 to the video decoding device 2, and the setting information needs to be embedded in the header. Nor.

When both blocks BK _A and BK _B are not completely in the screen (step S16: N), the difference vector code estimation units 13 and 27 subsequently indicate whether the blocks BK _A and BK _B are both completely out of the screen. Is determined (step S18).

When both blocks BK _A and BK _B are completely off the screen (step S18: Y), the difference vector code estimation units 13 and 27 indicate that "the distance between the block BK _A and its closest screen edge". The "distance between the block BK _B and its closest screen edge" is compared (step S19).

When the distance between the block BK _A and its closest screen edge is larger than the distance between the block BK _B and its closest screen edge (step S19: Y), the difference vector code estimation units 13 and 27 perform the difference vector. It is determined as an estimated code = negative (step S21).

Further, when the distance between the block BK _A and its closest screen edge is smaller than the distance between the block BK _B and its closest screen edge (step S19: N), the difference vector code estimation units 13 and 27 It is determined as the difference vector estimation code = positive (step S20).

For example, as shown in FIG. 4A, assuming that the positive sign of the difference vector indicates the right direction and the negative sign of the difference vector indicates the left direction, the block BK _A corresponding to the candidate vector _VA and the candidate vector V _B correspond to each other. An example in which both blocks BK _B are outside the screen will be described. When the distance _DA between the block BK _A and its closest screen edge is smaller than the distance _DB between the block BK _B corresponding to the candidate vector V _B and its closest screen edge, the difference vector estimation code = positive. By making a decision, it is determined that the block BK _A has a higher prediction accuracy than the block BK _B.

Further, when the distance between the block BK _A and its closest screen edge is equal to the distance between the block BK _B and its closest screen edge, the difference vector code estimation units 13 and 27 have the difference vector estimation code = positive. Alternatively, it is determined as the negative one specified in advance (difference vector estimation code = positive in step S19 shown in FIG. 3).

In the comparison according to step S19, the setting information shown as the second default value (or as the default value commonly used) is shown in the stream when the difference vector estimation code = positive or negative pre-specified one is determined. It can also be embedded in the header of (including one or more of a slice or tile header, a picture header, and a sequence header) for transmission. However, when the second default value is predetermined between the video coding device 1 and the video decoding device 2, it is not necessary to transmit the setting information from the video coding device 1 to the video decoding device 2, and the setting information is used as a header. There is no need to embed it in.

On the other hand, when both blocks BK _A and BK _B are neither completely in-screen nor off-screen (step S18: N), that is, when at least one of blocks BK _A and BK _B is partially off-screen. The difference vector code estimation units 13 and 27 calculate the area of the area outside the screen of the blocks BK _A and BK _B (step S22).
Area _SA = Area that protrudes outside the screen of block BK _A Area _SB = Area that protrudes outside the screen of block BK _B

Subsequently, the difference vector code estimation units 13 and 27 _compare the area _SA and the area SB (step S23).

When the area S _A is larger than the area _SB (step S23: Y), the difference vector code estimation units 13 and 27 determine that the difference vector estimation code = negative (step S25).

When the area S _A is smaller than the area _SB (step S23: N), the difference vector code estimation units 13 and 27 determine that the difference vector estimation code = positive (step S24).

For example, as shown in FIG. 4B, assuming that the positive sign of the difference vector indicates the right direction and the negative sign of the difference vector indicates the left direction, the block BK _A corresponding to the candidate vector _VA and the candidate vector V _B correspond to each other. An example in which a part of both the block BK _B and the block BK B is outside the screen will be described. When the area _SB is larger than the area _{SA, it is determined that the difference vector estimation code = positive, so that the block BK A has} a higher prediction accuracy than the block BK _B.

When the area _{S A} is equal to the area SB, the difference vector code estimation units 13 and 27 have the difference vector estimation code = positive or negative, whichever is specified in advance (in step S23 shown in FIG. 3, the difference vector estimation code = positive). ).

In the comparison according to step S23, the setting information shown as the third default value (or as the default value commonly used) is shown in the stream when the difference vector estimation code = positive or negative pre-designated one is determined. It can also be embedded in the header of (including one or more of a slice or tile header, a picture header, and a sequence header) for transmission. However, when the third default value is predetermined between the video coding device 1 and the video decoding device 2, it is not necessary to transmit the setting information from the video coding device 1 to the video decoding device 2, and the setting information is used as a header. There is no need to embed it in.

The difference vector estimation code determined through steps S17, S20, S21, S24, and S25 is set to the correct / incorrect flag 14 on the video coding device 1 side shown in FIG. 1, and the difference vector code is determined on the video decoding device 2 side shown in FIG. It is output to the unit 28 (step S26).

In this way, the difference vector code estimation units 13 and 27 calculate the positive and negative candidate vectors _VA and _BB using the block information of interest, the absolute value of the difference vector, and the prediction vector. The positive and negative signs of the difference vector are estimated based on the predetermined rules classified according to the positional relationship between the positions of the blocks BK _A and BK _B corresponding to the candidate vectors _VA and _BB , respectively, and the screen edge, and the difference is obtained. Determine the vector estimation sign.

Then, on the video coding device 1 side, the correct / incorrect flag generation unit 14 compares the difference vector estimation code obtained from the difference vector code estimation unit 13 with the positive / negative code of the difference vector obtained from the image coding unit 10. , Generates a correct / incorrect flag indicating "true" if they are equal and "false" if they are different.

Further, on the video decoding device 2 side, the difference vector sign determination unit 28 determines the difference vector estimation code as it is as the positive / negative sign of the difference vector when the correctness flag indicates “true”, and when it indicates “false”. , The sign obtained by reversing the difference vector estimation code is determined as the positive / negative sign of the difference vector.

(Principle of the present invention)
Here, the principle related to the estimation processing of the positive and negative signs of the difference vector of the first embodiment will be described. 5 (a), (b), and (c) show the block BKn of interest and the candidate vector _VA , which are related to the estimation processing of the positive and negative signs of the difference vector of the first embodiment in the video decoding apparatus 2 of the embodiment according to the present invention. It is explanatory drawing about the block BK _A , BK _B corresponding to V _B and a candidate vector _VA , V _B , respectively.

In order to simplify the explanation of the principle of the present invention with reference to FIG. 5, the prediction vector is set to 0, and only the sign of the horizontal component of the difference vector is focused on, and the positive and negative signs of the difference vector are obtained by the video decoding device 2. The decoding operation and the effect of the present invention will be described. In the description shown in FIG. 5, it is assumed that the plus sign of the difference vector indicates the right direction and the minus sign of the difference vector indicates the left direction.

As shown in FIG. 5A, it is assumed that the block of interest BKn exists at the edge of the screen in the screen. It is assumed that the absolute value of the difference vector having a predetermined length is input (read) to the block of interest BKn.

As shown in FIG. 5B, first, the difference vector code estimation units 13 and 27 calculate the positive and negative candidate vectors _VA and _BB with respect to the block of interest BKn (FIG. 3: step S14). ..

As shown in FIG. 5C, next, the difference vector code estimation units 13 and 27 calculate the block BK _A and the block BK _B in which the block BKn of interest is moved to the candidate vector _VA and the candidate vector _BB , respectively. (FIG. 3: Step S15).

In the example shown in FIG. 5C, since the block BK _A is located inside the screen and the block BK _B is located outside the screen, the difference vector code estimation units 13 and 27 are located outside the screen of the blocks BK _A and BK _B. Areas S _A and S _B of the region are calculated (FIG. 3: step S22).

Then, the difference vector code estimation units 13 and 27 _compare the area _SA and the area SB (FIG. 3: step S23). In the example shown in FIG. 5 (c), the area _{SB (= the area of the block BK B) protruding outside the screen of the block BK B is larger} than the area _SA (= 0) protruding outside the screen of the block BK _A. The difference vector estimation code of the block of interest BKn is “positive” (FIG. 3: step S24).

On the video coding device 1 side, if the difference vector estimation code obtained by the estimation process by the difference vector code estimation unit 13 is equal to the actual positive / negative code of the difference vector, it is true, and if it is different, it is false. Generates a flag, performs context-adaptive reversible compression processing, and transmits it. For example, the video coding device 1 transmits a correct / incorrect flag indicating true to the video decoding device 2.

On the video decoding device 2 side, when the correctness flag indicates true, the difference vector estimation code is used as the positive / negative sign of the difference vector, and when indicating false, the code obtained by inverting the estimated positive / negative code is used as the positive / negative sign of the difference vector. Determined as a sign. For example, when the video coding device 1 transmits a correct / incorrect flag indicating true to the video decoding device 2, the video decoding device 2 uses the difference vector estimation code as the positive / negative code of the difference vector. Therefore, in this example with reference to FIG. 5, the video decoding apparatus 2 finally determines that the positive and negative signs of the difference vector of the block of interest BKn are positive.

As described above, in the video coding device 1 and the video decoding device 2 according to the present invention, the positive and negative signs of the difference vector are represented by the correct / incorrect flags, and are compressed by the context-adaptive lossless compression for the following two reasons. Is possible.

The first reason is that, in general, when the block indicated by the candidate vector protrudes outside the screen, the effect of motion compensation prediction is reduced if the area outside the screen is large. For this reason, many motion vectors often do not point off the screen. Therefore, in the "estimation process of the positive / negative sign of the difference vector" according to the present invention, when the positive / negative sign of the difference vector of the block BKn of interest is assumed to be "positive" (candidate vector _VA , block BK _A ), "negative". Assuming that (candidate vector V _B , block BK _B ) is compared, the code with the smaller area outside the screen of each block BK _A and BK _B is used as the estimation code. Therefore, the probability that the correct / incorrect flag indicating the correctness of the estimated code becomes "true" increases, and compression (lossless compression) becomes possible.

The second reason is that compression (lossless compression) is possible even in the case of a scene where the assumption that "many motion vectors do not point out of the screen" in the first reason above does not hold. An example is shown in FIG.

FIG. 6 is an example of an image in which the camera is panned, and shows a state in which the frame F _N of interest is shifted to the left as a whole relative to the front frame F _N-1 with respect to the subject. For the focus block BKn located at the edge of the screen in the focus frame F _N , for example, the appropriate block as a reference destination for the motion compensation prediction of the focus block BKn is the block BK _A in the previous frame F _N-1 . be. That is, the motion vector of the block of interest BKn points out of the screen. For the sake of simplicity, if the prediction vectors are all 0, the positive and negative signs of the difference vector will be positive.

In the "estimation process of the positive / negative sign of the difference vector" according to the present invention, the difference vector code estimation units 13 and 27 calculate the candidate vectors _VA and _BB on the positive side and the negative side (FIG. 3: step S14). The block BK _A and the block BK _B in which the block BKn of interest is moved to the candidate vector _VA and the candidate vector V _B are calculated (FIG. 3: step S15).

At this time, the block BK _A has an area outside the screen, whereas the block BK _B stays in the screen. Therefore, the difference vector code estimation units 13 and 27 and the difference vector code estimation units 13 and 27 are the block BK. The areas _SA and _SB of the area outside the screen of _A and BK _B are calculated (FIG. 3: step S22).

Subsequently, the difference vector code estimation units 13 and 27 _compare the area _SA and the area SB (FIG. 3: step S23). In the example shown in FIG. 6, since the area _SB outside the screen of the block BK _B is smaller than the area _{SA outside the screen of the block BK A ,} the difference vector estimation code of the block BKn of interest is “negative” (candidate). Vector V _B , block BK _B ) (FIG. 3: step S25).

Therefore, on the video coding apparatus 1 side, the difference vector estimation code obtained by the estimation process by the difference vector code estimation unit 13 is different from the actual positive / negative code of the difference vector, and therefore the correct / incorrect value is regarded as “false”. Generates a flag and performs context-adaptive reversible compression processing for transmission.

Then, on the video decoding device 2 side, since the correct / incorrect flag indicates "false", the difference vector estimation code indicating the similarly estimated "negative" is inverted, and finally the positive / negative of the difference vector of the block of interest BKn. The sign is determined to be "positive" (candidate vector _VA , block BK _A ).

In this way, the difference vector estimation code is finally determined to be accurate “positive” (candidate vector _VA , block BK _A ) as the positive / negative sign of the difference vector for the block _BKn of interest in the frame FN of interest. The estimation code is "negative" (candidate vector V _B , block BK _B ), which is the code with the smaller area outside the screen of each block BK _A , BK _B , and the correct / incorrect flag is "false".

Similarly, when the blocks BKn (1), BKn (2), BKn (3), and BKn (4) located above the focus block _BKn located at the edge of the screen in the focus frame FN are set as the focus block, Each motion vector points out of the screen, and the correct / incorrect flag is "false" in each case.

Then, the video coding device 1 and the video decoding device 2 perform context-applied lossless compression / decoding processing on the correct / incorrect flags related to the positive / negative codes of the difference vector related to the block BKn of interest. In the context-applied lossless compression / decryption processing, the value around the block of interest BKn (for example, according to a predetermined context model such as the block directly above, the upper right, and the upper left) (that is, the value of the correctness flag) is adaptively applied. Lossless compression coding / decoding is performed by determining the bit value. In the example shown in FIG. 6, since a large number of blocks of correct / incorrect flags that are “false” are lined up in the periphery, the effect of compression is higher when the correct / incorrect flag is “false”, and compression (lossless compression) is possible. In particular, on the video decoding device 2 side, the context-applied reversible decoding process can restore the fact that the correct / incorrect flag is “false” for the block BKn of interest, and finally an accurate “correct” (candidate vector _VA , Block BK _A ) can be determined.

Therefore, according to the video coding device 1 and the video decoding device 2 according to the present invention, the lossless compression of the correct / incorrect flag becomes possible by increasing the probability that the correct / incorrect flag becomes "true", or a large number of "false" are made. When "" appears, there is a tendency for "false" blocks to gather around it, so it is possible to compress the correct / incorrect flag by context-adaptive lossless compression.

In particular, in the estimation process of the positive / negative sign of the difference vector in the first embodiment shown in FIG. Since it is equivalent to using (comparison of _SA and _SB ), it is effective from the viewpoint of the compression effect if it is applied only to the blocks located around the edge of the screen.

(Area where the correct / incorrect flag is used in the frame image)
As an embodiment of the video coding device 1 and the video decoding device 2 of the present embodiment, the setting control unit 18 in the video coding device 1 and the setting control unit 24 in the video decoding device 2 are the blocks of interest in the frame image. Depending on the position, the area where the correctness flag is used and the area where the positive / negative sign +/- is expressed as 1/0 bit as in the conventional technique can be controlled based on the predetermined setting information.

Then, as described above, in the video coding device 1 and the video decoding device 2 of the present embodiment, the correct / incorrect flag after the lossless compression or the simply binarized positive / negative code can be switched based on the setting information. can.

Hereinafter, more specifically, as an embodiment, the setting control unit 18 in the video coding device 1 and the setting control unit 24 in the video decoding device 2 have the attention block BKn from the area AR1 to the screen edge according to the setting information. The operation of switching the correct / incorrect flag or the simply binarized positive / negative code for the expression of the positive / negative code of the difference vector will be described based on whether or not it is contained in any of AR8.

FIG. 7 shows a region using the correct / incorrect flag according to the video coding device 1 and the video decoding device 2 of the embodiment according to the present invention, and +/- of the positive / negative code as a 1/0 bit representation as in the conventional technique. It is explanatory drawing of the frame image which has the area to perform.

As shown in FIG. 7, the frame F _N constituting the frame image can be separated into a region AR0 in the center of the screen and regions AR1 to AR8 within a predetermined range at the edge of the screen. The region AR0 is a region in which +/- of a positive / negative sign is represented by 1/0 as in the conventional technique. The areas AR1 to AR8 at the edge of the screen are areas for performing context-applied lossless compression / decryption processing of the correct / incorrect flag. Further, a different context model is set for each of the areas AR1 to AR8 at the edge of the screen.

These areas have widths D1, D2, D3, and D4 that are variable or fixed according to the size of the frame _FN and the size of the basic unit (block called a macro block or CTU) for encoding processing. It can be specified by the designation of, and it can be determined in advance between the video coding device 1 and the video decoding device 2, or in the form of transmitting the setting information, the configuration can be specified by the setting information. Further, D1, D2, D3, and D4 may have different values or may have the same value. The widths D1, D2, D3, and D4 are for the size of the basic unit (block called macro block, CTU, etc.) for encoding processing as an area where correct / incorrect flags indicating "true" or "false" are likely to be collected. It is preferable that the width is 1 to 3 times.

Since the difference vector is a two-dimensional vector, it is represented by two vectors, a horizontal component and a vertical component. Therefore, in the setting control unit 18 in the video coding device 1 and the setting control unit 24 in the video decoding device 2, as an embodiment of the setting information, the block BKn of interest fits in any of the areas AR1 to AR8 at the edge of the screen. It is possible to switch the correct / incorrect flag after the lossless compression or the simply binarized positive / negative code for each component of the difference vector depending on whether or not the value is satisfied.

For example, when the target block BKn is in the regions AR3 and AR4 near the left and right sides of the screen, the setting control unit 18 in the video coding apparatus 1 uses the correct / incorrect flag only for the horizontal component of the difference vector and performs context-adaptive reversible compression processing. Is performed, and the positive and negative codes in the vertical direction are controlled so that the +/- of the positive and negative codes are represented by 1/0 as in the conventional technique.

Further, in the setting control unit 18 in the video coding device 1, the setting control unit 18 in the video coding device 1 has only the vertical component of the difference vector when the block BKn of interest is in the regions AR1 and AR2 near the upper and lower sides of the screen. A context-adaptive reversible compression process is performed using the correct / incorrect flag, and the positive / negative code in the horizontal direction is controlled so that the +/- of the positive / negative code is represented by 1/0 as in the conventional technique.

Further, the setting control unit 18 in the video coding apparatus 1 uses correct / incorrect flags for both the horizontal component and the vertical component when the block BKn of interest is in the regions AR5, AR6, AR7, AR8 at the four corners of the screen, and is context-adaptive. Control to perform lossless compression processing.

Similarly, in the setting control unit 24 of the video decoding device 2, the lossless compression is performed for each component of the difference vector based on whether or not the block BKn of interest is contained in any of the areas AR1 to AR8 at the edge of the screen. It is possible to switch between the later correct / incorrect flag or the simply binarized positive / negative code.

For example, with reference to FIG. 8, the setting control unit 24 in the video decoding device 2 will explain the process of determining the positive / negative sign of the difference vector of one embodiment. FIG. 8 is a flowchart showing a positive / negative sign determination process of a difference vector of an embodiment in the video decoding apparatus 2 of the embodiment according to the present invention.

First, in the setting control unit 24, the block BKn of interest is contained within a predetermined range at the left and right edges of the screen (that is, within the regions AR3, AR4, AR5, AR6, AR7, AR8 near the left and right sides of the screen shown in FIG. 7). (Step S31).

Then, when the block BKn of interest is not within the predetermined range (within the regions AR3, AR4, AR5, AR6, AR7, AR8) at the left and right edges of the screen (step S31: N), the simple binary value is the same as in the conventional technique. The positive / negative code of the horizontal component of the difference vector is determined so as to be the converted positive / negative code (step S32).

On the other hand, when the block BKn of interest is within a predetermined range (within the regions AR3, AR4, AR5, AR6, AR7, AR8) at the left and right edges of the screen (step S31: Y), the "positive / negative sign of the difference vector" according to the present invention. The positive / negative sign of the horizontal component of the difference vector is estimated by the "estimation process", and the positive / negative sign of the horizontal component of the difference vector is determined using the correct / incorrect flag (step S33).

Subsequently, in the setting control unit 24, the block BKn of interest is contained within a predetermined range at the upper and lower ends of the screen (that is, within the regions AR1, AR2, AR5, AR6, AR7, AR8 near the left and right sides of the screen shown in FIG. 7). (Step S34).

Then, when the block BKn of interest is not within the predetermined range (within the regions AR3, AR4, AR5, AR6, AR7, AR8) at the left and right edges of the screen (step S34: N), the simple binary value is the same as in the conventional technique. The positive / negative code of the horizontal component of the difference vector is determined so as to be the converted positive / negative code (step S35).

On the other hand, when the block BKn of interest is within a predetermined range (within the regions AR3, AR4, AR5, AR6, AR7, AR8) at the left and right edges of the screen (step S34: Y), the “positive / negative sign of the difference vector” according to the present invention. The positive / negative sign of the horizontal component of the difference vector is estimated by the "estimation process", and the positive / negative sign of the horizontal component of the difference vector is determined using the correct / incorrect flag (step S36).

Therefore, in the case of a block that uses the correct / incorrect flag for only one of the horizontal component and the vertical component of the difference vector, one correct / incorrect flag is read from the bit stream, and if the block uses the correct / incorrect flag for both the horizontal component and the vertical component. Will read two correct / incorrect flags.

In the example shown in FIG. 8, an example of controlling so as to determine the positive / negative sign of the vertical component of the difference vector after determining the positive / negative sign of the horizontal component of the difference vector has been described. Later, it may be configured to control so as to determine the positive and negative signs of the horizontal component.

In this way, based on whether the block BKn of interest is contained in any of the areas AR1 to AR8 at the edge of the screen, the component using the correctness flag is switched between the horizontal component and the vertical component of the difference vector. be able to. For each of the horizontal component and the vertical component of the difference vector, the component that uses the correctness flag is predetermined between the video coding device 1 and the video decoding device 2, or is included in the setting information in the stream. Can be embedded in a header (including any one or more of a slice or tile header, a picture header, and a sequence header) and transmitted from the video coding device 1 to the video decoding device 2.

When the focus block BKn is in the regions AR1 to AR8 at the edge of the screen, the setting information is set so as to specify the determination order of the horizontal or vertical component of the difference vector according to the position of the focus block BKn. You can also keep it. This can uniquely determine the component in which the correct / incorrect flag is used according to the position of the block of interest BKn in relation to the switching of the correct / incorrect flag after the lossless compression or the simply binarized positive / negative sign, and sequentially. It is possible to omit extra processing such as determining which one is used, and there is an advantage that the processing operation becomes faster.

For example, in the setting control unit 18 in the video coding device 1 and the setting control unit 24 in the video decoding device 2, the block BKn of interest is placed in the areas AR1 to AR8 at the edge of the screen as the process of estimating the positive and negative codes of the difference vector. If it is closer to the left and right sides than the top and bottom sides of the screen when it is settled, it is controlled to first determine the positive and negative signs of the horizontal component of the difference vector and then determine the positive and negative codes of the vertical component.

Further, in the setting control unit 18 in the video coding device 1 and the setting control unit 24 in the video decoding device 2, the block BKn of interest is placed in the areas AR1 to AR8 at the edge of the screen as the process of estimating the positive and negative codes of the difference vector. When the block BKn of interest is closer to the upper and lower sides than the left and right sides of the screen when it is settled, the positive and negative codes of the vertical components of the difference vector are first determined, and then the positive and negative codes of the horizontal components are controlled to be determined.

Further, in the setting control unit 18 in the video coding device 1 and the setting control unit 24 in the video decoding device 2, the block BKn of interest is placed in the areas AR1 to AR8 at the edge of the screen as the process of estimating the positive and negative signs of the difference vector. If the blocks of interest are at equal distances from the left and right sides and the top and bottom sides of the screen when they are settled, the sign of the vertical component is specified after the code of the horizontal component is first obtained in a predetermined order (for example, the code of the horizontal component is first obtained. (Determining) controls to determine the positive and negative signs of the horizontal and vertical components of the difference vector.

Further, as a further modification, instead of the predetermined range at the screen edge, the correct / incorrect flag after the lossless compression or the simple It is possible to switch between the positive and negative codes binarized to. In particular, AR1 if the block BKn of interest touches only the upper side of the screen, AR2 if it touches only the lower side of the screen, AR3 if it touches only the left side of the screen, and AR4 if it touches only the right side of the screen. AR5 if it touches both the top and left sides of the screen, AR6 if it touches both the top and right sides of the screen, AR7 if it touches both the bottom and left sides of the screen, touches both the bottom and right sides of the screen. If it is, it is preferable to control it as a block that is contained in the area of AR8, and in other cases, it is preferable to control it as a block that is contained in the area of AR0.

(Block referenced as context)
As described above, in the video coding device 1 and the video decoding device 2 of the present embodiment, depending on the position of the block of interest in the frame image, the area where the correct / incorrect flag is used and the + of the positive / negative code as in the conventional technique. The area where /-is a 1/0 bit representation can be controlled based on predetermined setting information. Then, a context-adaptive lossless compression / decryption process is applied to the transmission of the correct / incorrect flag.

That is, as described above, the context-adaptive lossless compression unit 15 in the video coding device 1 and the context-adaptive lossless decoding unit 26 in the video decoding device 2 specify the context model based on the block information of interest. Then, a context-adaptive lossless compression coding / decoding process is performed on the correct / incorrect flag.

In the context-adaptive lossless compression coding / decoding process, for example, in the case of a video coding method according to the MPEG-H HEVC / H.265 method, CABAC is used as a video coding method according to the AVC / H.264 method. In some cases, CABAC or CAVLC can be used.

However, in the context model related to the lossless compression / decoding of the correct / incorrect flag according to the present embodiment, there is a preferable example as to which block in the vicinity of the block of interest is referred to as the context.

That is, in the context model according to the present embodiment, the position of the block to be referred to as the context is determined in advance according to the position of the block of interest BKn in the screen. This can be expected to further improve the compression efficiency.

More specifically, as a block to be referred to as a context according to the position of the focus block BKn in the screen, when the focus block BKn is closer to the left and right sides of the screen than the top and bottom sides of the screen, the block directly above the focus block BKn is referred to. do. For example, as shown in FIG. 9A, when the focus block BKn is closer to the left side (distance d1) of the screen (distance d1) than the bottom side (distance d2) of the screen (d1 <d2), refer to the block RBK directly above the block BKn of interest. do.

Further, as a block to be referred to as a context according to the position of the attention block BKn in the screen, when the focus block BKn is closer to the upper and lower sides of the screen than the left and right sides of the screen, the block immediately to the left of the focus block BKn is referred to. For example, as shown in FIG. 9B, when the focus block BKn is closer to the bottom side (distance d2) of the screen (distance d2) than the left side of the screen (distance d1) (d1> d2), the block RBK immediately to the left of the focus block BKn is used. refer.

Further, as a block to be referred to as a context according to the position of the focus block BKn in the screen, when the focus block BKn is equidistant from the upper and lower sides of the screen than the left and right sides of the screen, the focus block BKn is directly left, directly upper left, and so on. Refer to one or more predetermined blocks directly above and directly above. For example, as shown in FIG. 9C, when the focus block BKn is equidistant from the left side of the screen (distance d1) and the bottom side of the screen (distance d2) (d1 = d2), the block directly to the left of the focus block BKn. Refer to one or more predetermined blocks among RBK1, the block RBK2 on the upper left, the block RBK3 on the upper right, and the block RBK4 on the upper right.

As described above, one or a plurality of blocks are referred to as the context related to the lossless compression / decryption processing, and are adapted according to the CABAC or CAVLC method according to the value of each referenced block (that is, the value of the correctness flag). Lossless compression / decoding can be performed.

(Example 2: Estimating processing of positive / negative sign of difference vector)
In the “positive / negative sign estimation process of the difference vector” of the first embodiment shown in FIG. 3, the difference vector code estimation units 13 and 27 calculate the positive and negative candidate vectors _VA and VB for the block _BKn of interest. , An example of calculating the block BK _A and the block BK _B corresponding to these and determining the difference vector estimation code based on the positional relationship and the areas _SA and _SB protruding outside the screen has been described.

As an example of this modification, it is possible to perform the estimation processing of the positive / negative sign of the difference vector using only the coordinates of the block, thereby making the processing faster and reducing the processing load. Hereinafter, with reference to FIG. 10, the processing load can be reduced and the speed can be further increased. FIG. 10 is a flowchart showing a positive / negative sign estimation process of the difference vector of the second embodiment in the video coding device 1 and the video decoding device 2 according to the present invention.

Here, the estimation process of the positive and negative signs of the horizontal component of the difference vector will be described.

First, the difference vector code estimation units 13 and 27 according to the second embodiment input (read) the absolute value of the difference vector for the block of interest (step S11), and the absolute value of the difference vector = 0, as in the first embodiment. It is determined whether or not it is (step S12).

When the absolute value of the difference vector = 0 (step S12: Y), the difference vector code estimation units 13 and 27 according to the second embodiment end the estimation process with no positive or negative sign of the difference vector, as in the first embodiment. (Step S13).

On the other hand, when the absolute value of the difference vector ≠ 0 (step S12: N), the difference vector code estimation units 13 and 27 according to the second embodiment are on the positive side and the negative side with respect to the block of interest, as in the first embodiment. Candidate vectors _VA and _BB of are calculated (step S14).
Candidate vector _VA = Prediction vector + Absolute value of difference vector Candidate vector VA = Prediction vector _- Absolute value of difference vector

Subsequently, the difference vector code estimation units 13 and 27 according to the second embodiment instead of calculating the blocks BK _A and BK _B corresponding to the positive and negative candidate vectors _VA and _BB in the first embodiment, respectively. , The coordinates x of the upper left pixel of the focus block BKn, the width w of the focus block BKn, and the width (screen width) W of the frame image are set (step S15').

Subsequently, the difference vector code estimation units 13 and 27 according to the second embodiment determine whether or _{not “x + VA ≧ 0, x + w + VA ≦ W, and x + V B ≧ 0, and x + w + V B ≦ W ”} is satisfied (step). S16'). This process is an alternative process corresponding to the determination of whether the blocks BK _A and BK _B in the first embodiment are completely in the screen (FIG. 3: step S16).

When “x + _VA ≧ 0, x + w + _VA ≦ W, and x + V _B ≧ 0, and x + w + V _B ≦ W” are satisfied (step S16': Y), the difference vector code estimation units 13 and 27 according to the second embodiment The positive and negative signs of the difference vector are estimated as default values, and the difference vector estimated sign = default value is determined (step S17). In this example, "positive" is used as the default value, but "negative" may be used as the default value.

When “x + _VA ≧ 0, x + w + _VA ≦ W, and x + V _B ≧ 0, and x + w + V _B ≦ W” are not satisfied (step S16': N), the difference vector code estimation units 13 and 27 according to the second embodiment Then, it is determined whether or not "" x + w + _VA ≤ 0 or x + _VA ≥ W "and" x + w + V _B ≤ 0 or x + V _B ≥ W "" is satisfied (step S18'). This process is an alternative process corresponding to the determination of whether the blocks BK _A and BK _B in the first embodiment are completely out of the screen (FIG. 3: step S18).

When "" x + w + _VA ≤ 0 or x + _VA ≥ W "and" x + w + V _B ≤ 0 or x + V _B ≥ W "" are satisfied (step S18': Y), the difference vector code estimation unit 13 according to the second embodiment. , 27 compare "min (abs (x + w + _VA ), abs (x + _{VA -} W))" with "min (abs (x + w + VA), abs (x + VA _- W)" (step S19'). .Abs () is a function for calculating the absolute value, and min () is a function for calculating the minimum value. This process is the "distance between the block BK _A and its nearest screen edge" in the first embodiment. , And, it is an alternative process corresponding to the comparison (FIG. 3: step S19) of "the distance between the block _BKB and its closest screen edge".

When "min (abs (x + w + V _A ), abs (x + V _A -W))" is larger than "min (abs (x + w + V _B ), abs (x + V _B -W)" (step S19': Y), Examples. The difference vector code estimation units 13 and 27 according to No. 2 determine that the difference vector estimation code = negative (step S21).

Further, when "min (abs (x + w + V _A ), abs (x + V _A -W))" is smaller than "min (abs (x + w + V _B ), abs (x + V _B -W)" (step S19': N), The difference vector code estimation units 13 and 27 according to the second embodiment determine that the difference vector estimation code = positive (step S20).

Further, when "min (abs (x + w + V _A ), abs (x + V _A -W))" is equal to "min (abs (x + w + V _B ), abs (x + V _B -W)", the difference vector according to the second embodiment. The code estimation units 13 and 27 determine that the difference vector estimation code = positive or negative (in the example shown in FIG. 10, the difference vector estimation code = positive).

In the comparison according to step S19', the setting information shown as the second default value (or as the commonly used default value) is streamed when the difference vector estimation code = positive or negative pre-specified one is determined. It can be embedded in the header inside (including one or more of a slice or tile header, a picture header, and a sequence header) for transmission. However, when the second default value is predetermined between the video coding device 1 and the video decoding device 2, it is not necessary to transmit the setting information from the video coding device 1 to the video decoding device 2, and the setting information is used as a header. There is no need to embed it in.

On the other hand, when both steps S16'and 17'are not satisfied, the difference vector code estimation units 13 and 27 according to the second embodiment have "max (abs (min (0, x + _VA ))), max (W, x + w + _VA ). )-W) "and" max (abs (min (0, x + V _B )), max (W, x + w + V _B ) -W) "are compared (step S23'). max () is a function to calculate the maximum value. This process is an alternative process corresponding to the comparison between the area _SA and the area _SB in the first embodiment (FIG. 3: step S23).

"Max (abs (min (0, x + _VA )), max (W, x + w + _VA ) -W)" is "max (abs (min (0, x + V _B )), max (W, x + w + V _B ) -W" ) ”(Step S23': Y), the difference vector code estimation units 13 and 27 according to the second embodiment determine that the difference vector estimation code = negative (step S25).

"Max (abs (min (0, x + _VA )), max (W, x + w + _VA ) -W)" is "max (abs (min (0, x + V _B )), max (W, x + w + V _B ) -W" ) ”(Step S23': N), the difference vector code estimation units 13 and 27 according to the second embodiment determine that the difference vector estimation code = positive (step S24).

"Max (abs (min (0, x + _VA )), max (W, x + w + _VA ) -W)" is "max (abs (min (0, x + _VA ))), max (W, x + w + VA) _-W . ) ”, The difference vector code estimation units 13 and 27 according to the second embodiment assume that the difference vector estimation code = positive or negative, whichever is specified in advance (in the example shown in FIG. 10, the difference vector estimation code = positive). decide.

In the comparison according to step S23', the setting information shown as the third default value (or as the commonly used default value) is streamed when the difference vector estimation code = positive or negative pre-specified one is determined. It can be embedded in the header inside (including one or more of a slice or tile header, a picture header, and a sequence header) for transmission. However, when the third default value is predetermined between the video coding device 1 and the video decoding device 2, it is not necessary to transmit the setting information from the video coding device 1 to the video decoding device 2, and the setting information is used as a header. There is no need to embed it in.

The difference vector estimation code determined through steps S17, S20, S21, S24, and S25 is set to the correct / incorrect flag 14 on the video coding device 1 side shown in FIG. 1, and the difference vector code is determined on the video decoding device 2 side shown in FIG. It is output to the unit 28 (step S26).

In this way, the difference vector code estimation units 13 and 27 according to the second embodiment calculate the positive and negative candidate vectors _VA and _BB using the absolute value of the difference vector and the prediction vector. The positive and negative signs of the difference vector are estimated based on the predetermined rules classified according to the coordinates and size of the block BKn of interest, the candidate vectors _VA , _BB , and the positional relationship of the screen width, and the difference vector estimation code is obtained. decide. Then, as compared with the first embodiment, in the second embodiment, the coordinate calculation is sufficient instead of obtaining the area, so that the rule can be simplified. In addition, in FIG. 10, instead of using the functions abs (), min () and max (), the if statement may be used for classification.

Then, on the video coding device 1 side, the correct / incorrect flag generation unit 14 compares the difference vector estimation code obtained from the difference vector code estimation unit 13 with the positive / negative code of the difference vector obtained from the image coding unit 10. , Generates a correct / incorrect flag indicating "true" if they are equal and "false" if they are different.

Further, on the video decoding device 2 side, the difference vector sign determination unit 28 determines the difference vector estimation code as it is as the positive / negative sign of the difference vector when the correctness flag indicates “true”, and when it indicates “false”. , The sign obtained by inverting the difference vector estimation sign is determined as the positive / negative sign of the difference vector.

Therefore, the video coding device 1 and the video decoding device 2 each including the difference vector code estimation units 13 and 27 according to the second embodiment are video coding devices including the difference vector code estimation units 13 and 27 according to the first embodiment, respectively. The same actions and effects as those of 1 and the video decoding device 2 can be obtained.

(Example 3: Estimating processing of positive / negative sign of difference vector)
As a further modification of the second embodiment, as shown in FIG. 11, steps S16', S18', and S19' in FIG. 10 are skipped, and step S23'is executed immediately after step S15'to simplify the process. However, in many cases, the same effect can be obtained as lossless compression / decoding with respect to the positive and negative codes of the difference vector.

The video coding device 1 or the video decoding device 2 in each of the above embodiments can be configured by a computer, and a program for functioning each processing unit of the video coding device 1 or the video decoding device 2 is suitable. Can be used for. Specifically, the control unit for controlling each processing unit of the video coding device 1 or the video decoding device 2 can be configured by the central processing unit (CPU) in the computer, and each processing unit is operated. A storage unit for appropriately storing a program required for the above can be configured with at least one memory. That is, by causing such a computer to execute the program by the CPU, it is possible to realize the functions of each processing unit of the video coding device 1 or the video decoding device 2. Further, a program for realizing the functions of each processing unit of the video coding device 1 or the video decoding device 2 can be stored in a predetermined area of the above-mentioned storage unit (memory). Such a storage unit can be configured by a RAM or ROM inside the device, or can be configured by an external storage device (for example, a hard disk). Further, such a program can be configured as a part of software (stored in ROM or an external storage device) on an OS used by a computer. Further, a program for causing such a computer to function as each processing unit of the video coding device 1 or the video decoding device 2 can be recorded on a computer-readable recording medium. Further, each processing unit of the video coding device 1 or the video decoding device 2 may be configured as a part of hardware or software, and each may be combined and realized.

Although the present invention has been described above with reference to examples of specific embodiments, the present invention is not limited to the examples of the above-described embodiments, and can be variously modified without departing from the technical idea. For example, in the example of the above-described embodiment, the rules regarding the difference vector code estimation process have been described by exemplifying a specific order / calculation method, but the order may be changed, a simpler arithmetic expression (if statement, etc.), or more. Various modifications such as high-precision arithmetic expressions (combination of coordinates and areas, etc.) are assumed.

Further, in the example of the above-described embodiment, the center of the screen (FIG. 7: region AR0) is subjected to binarization processing in which +/- of the positive / negative sign of the conventional technique is represented by 1/0, and switching is performed. However, the center of the screen (FIG. 7: region AR0) may also be compressed transmission using the correct / incorrect flag according to the present invention. That is, it is assumed that the probability of occurrence is unlikely to be biased even when compressed transmission using the correct / incorrect flag in the center of the screen (Fig. 7: region AR0), so improvement in compression efficiency may not be expected. In other words, there is no significant deterioration. Instead, there are advantages such as simplification of processing (reduction of mounting capacity) in that the processing is unified with respect to the transmission of the positive and negative codes of the difference vector. Therefore, the video coding device and the video decoding device according to the present invention are not limited to the examples of the above-described embodiments, but are limited only by the description of the scope of claims.

According to the present invention, lossless compression is possible with respect to the positive and negative codes (+,-) of the difference vector in the video coding method using motion compensation prediction, and the compression efficiency is improved. Therefore, the video coding device and the video decoding Useful for device applications.

1 Video coding device 10 Image coding unit 11 Motion compensation prediction unit 12 Difference vector generation unit 13 Difference vector code estimation unit 14 Correct / wrong flag generation unit 15 Context-adaptive lossless compression unit 16 Positive / negative binarization unit 17 Switching unit 18 Setting control Part 19 Stream generation part 20 Stream reading part 21 Image decoding part 22 Difference vector restoration part 23 Motion compensation prediction part 24 Setting control part 25 Switching part 26 Context-adaptive lossless decoding part 27 Difference vector code estimation part 28 Difference vector code determination part 29 Positive / negative decision section

Claims (10)

It is a video coding device in a video coding method using motion compensation prediction.
A difference vector generation means that generates a difference vector based on motion compensation prediction for a predetermined block in a frame,
The estimation code of the difference vector is obtained by estimating the positive and negative signs of the difference vector according to a predetermined rule based on the information about the predetermined block, the absolute value of the difference vector, and the prediction vector related to the difference vector. Difference vector sign estimation means to determine and
The positive / negative sign of the difference vector obtained from the difference vector generation means is compared with the estimation code of the difference vector obtained from the difference vector code estimation means, and if they are equal, it indicates true, and if they are different, it indicates false. Correct / incorrect flag generation means for generating correct / incorrect flags, and
A context-adaptive lossless compression means that performs a lossless compression process adapted to the context according to the position of the predetermined block on the correctness flag generated by the correctness flag generation means.
A stream generation means for generating a stream for transmitting the correct / incorrect flag subjected to the lossless compression process in association with the absolute value of the difference vector.
A video coding device characterized by comprising. The difference vector code estimation means calculates positive and negative candidate vectors using the information about the predetermined block, the absolute value of the difference vector, and the prediction vector, and blocks corresponding to each of the candidate vectors. Depending on the positional relationship between the position of the frame and the screen edge of the frame, or according to the coordinate and size of the predetermined block, the positive and negative candidate vectors, and the positional relationship of the screen width of the frame. The video coding apparatus according to claim 1, wherein the positive / negative sign of the difference vector is estimated based on a predetermined rule divided into cases. The frame is divided into an area at the center of the screen and an area within a predetermined range at the edge of the screen, and the correct / incorrect flag generation means is activated when the predetermined block is contained in at least the area within the predetermined range at the edge of the screen. The video coding apparatus according to claim 1 or 2, further comprising a control means for controlling such. The control means is correct or incorrect for each of the horizontal component and the vertical component of the difference vector, based on which region of the predetermined range of the screen edge is contained in the predetermined block. The video coding apparatus according to claim 3, further comprising a means for controlling to switch the component using the flag. The control means is
When the predetermined block is closer to the left and right sides than the upper and lower sides of the screen of the frame, the difference vector code estimation means is made to determine the positive and negative signs of the horizontal component of the difference vector first.
When the predetermined block is closer to the upper and lower sides than the left and right sides of the screen of the frame, the difference vector code estimation means is made to determine the positive and negative signs of the vertical component of the difference vector first.
When the predetermined block is equidistant from the left and right sides and the top and bottom sides of the screen of the frame, the difference vector code estimation means is made to determine the positive and negative signs of the horizontal component and the vertical component of the difference vector in a predetermined order. The video coding apparatus according to claim 4, further comprising means for controlling such. The stream transmitted by the video coding apparatus according to any one of claims 1 to 5 is input and read, and the absolute value of the difference vector and the lossless compression process are performed for a predetermined block in the frame to be decoded. A stream reading means to extract the correctness flag applied, and
The estimation code of the difference vector is obtained by estimating the positive and negative signs of the difference vector according to a predetermined rule based on the information about the predetermined block, the absolute value of the difference vector, and the prediction vector related to the difference vector. Difference vector sign estimation means to determine and
A context-adaptive reversible decoding means that generates a deciphered correct / incorrect flag by performing a reversible decoding process adapted to the context according to the position of the predetermined block on the correct / incorrect flag subjected to the lossless compression process.
When the decoded correct / incorrect flag indicates true, the estimated sign of the difference vector is used as the positive / negative sign of the difference vector of the predetermined block, and when indicating false, the code obtained by inverting the estimated code of the difference vector is used as the difference. The difference vector sign determining means for determining the positive and negative signs of the vector,
In order to restore the image of the predetermined block, the absolute value of the extracted difference vector and the positive / negative sign of the difference vector determined by the difference vector sign determining means are used to determine the predetermined value in the frame to be decoded. Difference vector restoration means to restore the difference vector of the block of
A video decoding device characterized by comprising. The context-adaptive lossless decoding means refers to a block at a predetermined position as a block to be referred to as a context according to the position of the predetermined block in the screen, and is subjected to the lossless compression process. The video decoding device according to claim 6, wherein the lossless decoding process is performed on the device. The context-adaptive reversible decoding means is used as a block to be referred to as a context according to the position of the predetermined block in the screen.
If the predetermined block is closer to the left and right sides of the screen than the top and bottom sides of the screen, refer to the block directly above the predetermined block.
If the predetermined block is closer to the top and bottom sides of the screen than the left and right sides of the screen, refer to the block directly to the left of the predetermined block.
When the predetermined block is equidistant from the upper and lower sides of the screen than the left and right sides of the screen, refer to one or more predetermined blocks of the predetermined block directly left, directly upper left, directly above, and directly upper right. The video decoding device according to claim 7, wherein the lossless decoding process is performed on the correct / incorrect flag to which the lossless compression process has been performed. A program for operating a computer as the video coding apparatus according to any one of claims 1 to 5. A program for operating a computer as the video decoding device according to any one of claims 6 to 8.

Images