JP2017127022A - Video decoder, video decoding method, and video decoding program, and receiver, reception method and reception program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To solve such a problem that a prediction motion vector unrelated to the original motion of a decoding object block is derived, when correlation between the decoding object block and a peripheral block is low.SOLUTION: A motion block decoding unit 550 derives the motion vector of a decoding object prediction block, based on the motion vector of any one of the candidate blocks selected from peripheral blocks. A motion compensation prediction unit 530 performs motion compensation prediction using a motion vector thus derived. A motion vector decoding unit 550 determines whether or not a second block is set as a candidate block based on whether or not the number of motion vectors of the first block is same as that of the second block, and a reference index indicating a reference picture referred to by the motion vectors of the first block is same as that referred to by the motion vectors of the second block.SELECTED DRAWING: Figure 7

Description

The present invention relates to a video decoding technique using motion compensated prediction, and more particularly to a technique for deriving a motion vector used in motion compensated prediction.

In video compression coding, motion compensation prediction is generally used in order to increase compression efficiency by utilizing temporal correlation between pictures. In motion compensation prediction, a prediction value is obtained from a reference picture in consideration of the direction and amount of motion indicated by a motion vector. By encoding the difference value between the prediction value obtained by motion compensation prediction and the pixel value of the encoding target picture, redundancy can be removed and compression efficiency can be increased.

In MPEG-4AVC, which is the latest video compression coding, one picture can be I picture for intra-picture coding, P picture capable of unidirectional motion compensation prediction, unidirectional or bidirectional motion compensation prediction Encoded as a B picture.

When coding a motion vector of a certain coding target block, first, a predicted motion vector is derived from neighboring blocks. Then, a differential motion vector between the motion vector of the encoding target block and the predicted motion vector is calculated, and the differential motion vector is entropy encoded. Here, the reason why the motion vector predictor is derived from the peripheral block is that the motion vector of the encoding target block is considered to have a correlation with the motion vector of the peripheral block. In entropy coding, this correlation is used to perform compression so that the motion vector compression efficiency increases as the absolute value of the difference motion vector decreases.

Patent Document 1 discloses a motion vector prediction method using peripheral blocks of an encoding target block.

JP 2004-96705 A

In the above-described conventional motion vector encoding method, since the prediction motion vector is derived for each prediction direction such as forward and backward, the motion of an object having a temporal correlation different from the motion of the block to be encoded is a neighboring block. If it exists in the block, a motion vector predictor unrelated to the original motion of the object in the encoding target block is derived, and the compression efficiency of the motion vector may not be improved.

The present invention has been made in view of such a situation, and an object of the present invention is to provide a motion that can improve the accuracy of motion vector derivation on the decoding side while improving the compression efficiency of the motion vector on the encoding side. It is to provide a vector encoding technique and a motion vector decoding technique.

A moving picture decoding apparatus that decodes a code string encoded using motion compensated prediction in units of blocks obtained by dividing each picture of a moving picture, wherein a motion vector of a decoding target prediction block is obtained around the decoding target prediction block A motion information deriving unit derived based on a motion vector of any candidate block of candidate blocks selected from the blocks, and the decoding target prediction by motion compensated prediction using the motion vector derived by the motion information deriving unit A motion compensation prediction unit that generates a prediction signal of the block, and the motion information deriving unit includes the first block when the number of motion vectors of the first block and the number of motion vectors of the second block are two. A reference index indicating a reference picture referred to by the first motion vector of the first reference and a reference referred to by the first motion vector of the second block A reference index indicating a reference picture indicating a reference picture referenced by a second motion vector of the first block and a reference picture referred to by a second motion vector of the second block are the same. It is determined whether or not the second block is a candidate block depending on whether or not the reference indexes are the same, and the peripheral block includes at least a block adjacent to the upper left of the decoding target prediction block. A moving picture decoding apparatus is provided.
A moving picture decoding method for decoding a code string encoded by using motion compensated prediction in units of blocks obtained by dividing each picture of a moving picture, wherein a motion vector of a decoding target prediction block is calculated around the decoding target prediction block The decoding target prediction by motion information deriving step derived based on the motion vector of any candidate block of candidate blocks selected from the blocks, and motion compensation prediction using the motion vector derived by the motion information deriving unit A motion compensation prediction step for generating a prediction signal of the block, wherein the motion information deriving step includes the first block when the number of motion vectors of the first block and the number of motion vectors of the second block are two. A reference index indicating a reference picture referred to by the first motion vector of the block and a first motion vector of the second block The reference index indicating the reference picture referenced by the second block is the same, and the reference index indicating the reference picture referenced by the second motion vector of the first block and the second motion vector of the second block are referenced. It is determined whether the second block is a candidate block according to whether the reference indexes indicating the reference pictures to be used are the same, and the neighboring block is a block adjacent to the upper left of the decoding target prediction block. There is provided a moving image decoding method characterized by comprising at least.
A moving picture decoding program for decoding a code string encoded by using motion compensated prediction in units of blocks obtained by dividing each picture of a moving picture, wherein a motion vector of a decoding target prediction block is obtained around the decoding target prediction block The decoding target prediction by motion information deriving step derived based on the motion vector of any candidate block of candidate blocks selected from the blocks, and motion compensation prediction using the motion vector derived by the motion information deriving unit A motion compensation prediction step for generating a prediction signal of the block, and the motion information deriving step includes a step of calculating the motion information when the number of motion vectors of the first block and the number of motion vectors of the second block is two. A reference index indicating a reference picture to which a first motion vector of one block refers, and the second The reference index indicating the reference picture to which the first motion vector of the lock refers is the same, and the reference index indicating the reference picture to which the second motion vector of the first block refers and the second index of the second block It is determined whether the second block is a candidate block according to whether the reference indexes indicating the reference pictures to which the two motion vectors refer are the same, and the peripheral block is the decoding target prediction block There is provided a moving picture decoding program characterized by including at least an adjacent block on the upper left.
A receiving device that decodes a code string encoded using motion compensated prediction in units of blocks obtained by dividing each picture of a moving image, the receiving unit receiving packetized encoded data via a network; A memory that buffers received encoded data; a packet processing unit that packet-processes the encoded data to generate an encoded stream; and a motion vector of a decoding target prediction block in the encoded stream, the decoding target prediction A motion information deriving unit derived based on a motion vector of any candidate block of candidate blocks selected from neighboring blocks of the block; and the motion compensation prediction using the motion vector derived by the motion information deriving unit. A motion compensation prediction unit that generates a prediction signal of a decoding target prediction block, and the motion information derivation unit includes: When the number of motion vectors of one block and the number of motion vectors of the second block are 2, a reference index indicating a reference picture referred to by the first motion vector of the first block and the second block of the second block A reference index indicating a reference picture referenced by one motion vector is the same, and a reference index indicating a reference picture referenced by a second motion vector of the first block and a second motion of the second block It is determined whether or not the second block is a candidate block depending on whether or not the reference index indicating the reference picture referred to by the vector is the same, and the neighboring block is adjacent to the upper left of the decoding target prediction block There is provided a receiving device characterized in that it includes at least a block.
A reception method for decoding a code string encoded using motion compensated prediction in units of blocks obtained by dividing each picture of a moving image, wherein a reception step of receiving packetized encoded data via a network; A step of buffering received encoded data; a packet processing step of packet-processing the encoded data to generate an encoded stream; and a motion vector of a decoding target prediction block in the encoded stream; The motion information deriving step derived based on the motion vector of any candidate block selected from the neighboring blocks of the block, and the motion compensation prediction using the motion vector derived by the motion information deriving unit A motion compensation prediction step for generating a prediction signal of the prediction block to be decoded. In the motion information deriving step, when the number of motion vectors of the first block and the number of motion vectors of the second block are 2, the reference indicating the reference picture to which the first motion vector of the first block refers A reference index indicating a reference picture referenced by a second motion vector of the first block, and an index and a reference index indicating a reference picture referenced by a first motion vector of the second block are the same; It is determined whether or not the second block is a candidate block according to whether or not the reference index indicating the reference picture referred to by the second motion vector of the second block is the same. There is provided a reception method including at least a block adjacent to the upper left of the decoding target prediction block.
A reception program for decoding a code string encoded using motion compensated prediction in units of blocks obtained by dividing each picture of a moving image, and receiving a packetized encoded data via a network; A step of buffering received encoded data; a packet processing step of packet-processing the encoded data to generate an encoded stream; and a motion vector of a decoding target prediction block in the encoded stream; The motion information deriving step derived based on the motion vector of any candidate block selected from the neighboring blocks of the block, and the motion compensation prediction using the motion vector derived by the motion information deriving unit Motion compensated prediction step for generating a prediction signal of a decoding target prediction block The motion information deriving step refers to the first motion vector of the first block when the number of motion vectors of the first block and the number of motion vectors of the second block are two. A reference index indicating a reference picture and a reference index indicating a reference picture referred to by the first motion vector of the second block are the same, and a reference picture referred to by the second motion vector of the first block is referred to. It is determined whether or not the second block is a candidate block depending on whether or not the reference index indicating the reference picture indicating the reference picture referred to by the second motion vector of the second block is the same. The peripheral block includes at least a block adjacent to the upper left of the decoding target prediction block. To provide a credit program.

It should be noted that any combination of the above-described constituent elements and the expression of the present invention converted between a method, an apparatus, a system, a recording medium, a computer program, etc. are also effective as an aspect of the present invention.

According to the present invention, by considering the motion vectors of the peripheral blocks, it is possible to improve the accuracy of motion vector derivation while improving the compression efficiency of motion vectors on the encoding side.

It is a figure explaining the relationship between an encoding object block and a surrounding block. It is a flowchart explaining the conventional motion vector prediction method. It is a figure explaining the specific example of the conventional motion vector prediction method. It is a figure which shows the structure of the moving image encoder of 1st Embodiment. FIG. 5 is a diagram illustrating a detailed configuration of a motion vector number acquisition unit and a motion vector encoding unit in FIG. 4. It is a flowchart explaining the detail of the derivation | leading-out method of the prediction motion vector of 1st Embodiment. It is a figure which shows the structure of the moving image decoding apparatus of 1st Embodiment. It is a figure explaining the detailed structure of the motion vector number acquisition part of FIG. 7, and a motion vector decoding part. It is a figure which shows the structure of the moving image encoder of 2nd Embodiment. It is a flowchart explaining the detail of the derivation | leading-out method of the prediction motion vector of 2nd Embodiment. It is a figure explaining the relationship between a macroblock boundary and a prediction block. It is a figure which shows the structure of the moving image decoding apparatus of 2nd Embodiment. It is a figure explaining another specific example of the conventional motion vector prediction method. It is a flowchart explaining the detail of the derivation | leading-out method of the prediction motion vector of 3rd Embodiment.

  First, a technique that is a premise of the embodiment of the present invention will be described.

Currently, apparatuses and systems that comply with an encoding method such as MPEG (Moving Picture Experts Group) are widely used. In such an encoding method, a plurality of images that are continuous on the time axis are handled as digital signal information. At that time, for the purpose of broadcasting, transmitting or storing highly efficient information, compression using motion compensation prediction using temporal redundancy and orthogonal transform such as discrete cosine transform using spatial redundancy Encode.

In 1995, the MPEG-2 video (ISO / IEC 13818-2) encoding method was established as a general-purpose video compression encoding method, and a digital VTR of DVD (Digital Versatile Disk) and D-VHS (registered trademark) standards. Is widely used as an application for storage media such as magnetic tape and digital broadcasting.

Furthermore, in 2003, the Joint Technical Committee (ISO / IEC) of the International Organization for Standardization (ISO) and the International Electrotechnical Commission (IEC) and the International Telecommunication Union Telecommunication Standardization Department (IT)
A coding scheme called MPEG-4 AVC / H.264 (I
The standard number is 14946-10 for SO / IEC and H.264 for ITU-T. This is hereinafter referred to as MPEG-4 AVC) as an international standard.

In these encoding schemes, motion compensated prediction is used. In motion compensation prediction, a prediction value is obtained from a reference picture in consideration of the direction and amount of motion indicated by a motion vector. By encoding the difference value between the prediction value obtained by motion compensation prediction and the pixel value of the encoding target picture, redundancy can be removed and compression efficiency can be increased.

Pictures are divided into three types depending on how motion compensated prediction is used. These are an I picture that does not use motion compensated prediction, a P picture that can perform unidirectional motion compensated prediction, and a B picture that can perform unidirectional or bidirectional motion compensated prediction.

In the MPEG-2 P picture, motion compensation prediction is performed in units of macroblocks using the immediately preceding I picture or P picture in the display order as a reference picture. In contrast, MPEG-4
In AVC, a plurality of encoded pictures can be used as reference pictures, and motion compensation prediction can be performed by selecting an optimum picture for each prediction block (described later). In addition to the preceding picture in the display order, it is possible to refer to subsequent pictures in the display order.

Also, in the MPEG-2 B picture, motion compensation prediction using one reference picture in the display order, motion compensation prediction using one reference picture in the rear, or the two reference pictures Any of motion compensated prediction can be performed by averaging. On the contrary,
In MPEG-4AVC, an arbitrary reference picture can be used for motion compensation prediction regardless of the front or rear, regardless of the restriction of one front and one rear in the display order. Furthermore, a B picture can be referred to as a reference picture.

In this way, since any encoded picture can be referred to in MPEG-4 AVC, the decoded picture is stored and managed in the reference frame memory.

In MPEG-4 AVC, a reference picture list is designated by arranging pictures stored in a reference frame memory by a predetermined method to create a reference picture list and describing the index in the encoded stream. Here, the “reference picture list” is a list in which reference pictures in a reference frame used in motion compensation prediction can be rearranged. Coding efficiency can be improved by rearranging the reference pictures according to the frequency of use using the reference picture list.

The reference picture list can be rearranged by sending reference picture list change information in a slice header added for each slice.

In motion compensated prediction of MPEG-4 AVC, an encoded picture registered in the reference picture list is used as a reference picture for each prediction block, and a prediction value is calculated from the reference picture in consideration of the direction and amount of motion indicated by the motion vector. Is calculated.

The reference picture list is prepared in two directions. In general, a reference picture list L0 for registering a temporally forward-coded picture of an encoding target picture and, in general, a temporally backward-encoded picture of a encoding target picture is registered. Reference picture list L1
It is. The P picture uses the reference picture list L0, and the B picture uses the reference picture list L0 and the reference picture list L1.

Direction of motion compensation prediction by reference picture list L0 is direction 0, reference picture list L1
Let direction 1 be the direction of motion compensated prediction.

The prediction block will be described. First, a picture is divided into units of macro blocks (hereinafter referred to as MB) of 16 horizontal pixels × vertical 16 pixels, and the MB is further divided into 16 horizontal pixels × 16 vertical pixels, 16 horizontal pixels × 8 vertical pixels, The prediction block is divided into any of prediction blocks of horizontal 8 pixels × vertical 16 pixels, horizontal 8 pixels × vertical 8 pixels, horizontal 8 pixels × vertical 4 pixels, horizontal 4 pixels × vertical 8 pixels, and horizontal 4 pixels × vertical 4 pixels.

A motion vector indicating a motion direction and a motion amount and a reference index indicating a reference picture in a reference picture list are assigned to each prediction block of a P picture. One or two motion vectors indicating a motion direction and a motion amount and a reference index indicating a reference picture in a reference picture list are assigned to a prediction block of a B picture.

In a prediction block to which a motion vector and a reference index are assigned one by one (hereinafter referred to as unidirectional prediction), a pixel value of a reference picture indicated by the motion vector and the reference index is a prediction value. On the other hand, in a prediction block (hereinafter referred to as bi-directional prediction) in which two motion vectors and two reference indexes are assigned, an average value of pixel values of reference pictures indicated by the respective motion vectors and reference indexes is a predicted value.

Next, as a conventional method for deriving a motion vector predictor, a method for deriving a motion vector predictor of Patent Document 1 and MPEG-4 AVC will be described with reference to FIGS. Hereinafter, unless otherwise specified, a block indicates a prediction block.

As shown in FIG. 1A, block A, block B, and block C that are peripheral blocks of the encoding target block X are determined (step S5000). Here, the reference index in the direction 0 of the block A is defined as refIdxL0A, the motion vector in the direction 0 is defined as mvL0A, the reference index in the direction 1 is defined as refIdxL1A, and the motion vector in the direction 1 is defined as mvL1.
A is defined. Similarly, refIdxL0B, mvL0B, refId of block B
xL1B, mvL1B, block C refIdxL0C, mvL0C, refIdxL
Define 1C, mvL1C.

It is checked whether or not motion-compensated prediction in the direction 0 is performed in the encoding target block X (step S5010).

When motion compensation prediction in the direction 0 is not performed (NO in step S5010), the reference index refIdxL0X in the direction 0 of the encoding target block X is set to −1, and the motion vector mv.
L0X is set to (0, 0) (step S5030).

When motion compensation prediction in the direction 0 is performed (YES in step S5010), the reference indices refIdxL0A and refIdxL in the direction 0 of the three neighboring blocks A, B, and C are used.
It is checked whether only one of 0B and refIdxL0C is the same as refIdxL0X, which is the reference index of the encoding target block X (step S5020).
If there is only one peripheral block whose reference index matches the encoding target block X (YES in step S5020), the motion vector mvL0N of the peripheral block N is set as the predicted motion vector mvpL0X of the encoding target block X ( Step S5040).

Whether there is a neighboring block whose reference index matches the encoding target block X or 2
If there are two or more (NO in step S5020), as shown in Equation 1, the peripheral block A
The median value of each component of motion vectors mvL0A, mvL0B, and mvL0C in directions 0, B, and C is set as a predicted motion vector mvpL0X (step S5050).
mvpL0X (x) = Median [mvL0A (x), mvL0B (x), mvL0
C (x)]
mvpL0X (y) = Media [mvL0A (y), mvL0B (y), mvL0
C (y)] (Formula 1)

For the motion compensation prediction in direction 1 in the encoding target block X, processing similar to that in direction 0 is performed (steps S5110 to S5150).

The problems of the conventional method described above will be described. In a conventional motion vector prediction method, a predicted motion vector is prevented from taking an unexpected value by using a median filter.
However, the median filter has a problem that the most efficient motion vector cannot be used as a predicted value.

Therefore, in the conventional motion vector prediction method, when there is only one peripheral block having the same reference index as the reference index of the encoding target block for each prediction direction in the preceding stage of the median filter, the motion vector of the peripheral block is present. Has been introduced as a predicted motion vector.

In general, since the motion vector depends on the motion of the object, the motion of the object has a correlation in the time direction. However, in the conventional motion vector prediction method, whether or not the reference indexes of the encoding target block and the neighboring blocks match is independently verified for each prediction direction, and therefore, the temporal correlation different from the motion of the encoding target block. When there is a motion of an object having a symbol in a peripheral block, there is a problem that the original motion of the encoding target block cannot be predicted correctly.

With reference to FIG. 3, the problem of such a conventional motion vector prediction method will be described with a specific example.

Block A, block B around the encoding target block X of the encoding target picture,
The positional relationship of the block C is as illustrated. Since the same object exists in the encoding target block X and the block B, bi-directional prediction using the forward reference picture P0 and the backward reference picture P1 is used for the encoding target block X and the block B. On the other hand, the object including the block C has a strong correlation with the forward reference picture P0 and unidirectional prediction is used, and the object including the block A has a strong correlation with the backward reference picture P1 and uses the unidirectional prediction.

The encoding target block X performs motion compensation prediction with a motion vector of (−8, 8) for the reference picture P0, and motion compensation prediction with a motion vector of (8, −8) for the reference picture P1. I do. The block A has a motion vector (0, 0) with respect to the reference picture P1.
Motion compensation prediction is performed. The block B has a motion vector ((
−8,8), and the motion vector is (8,8) for the reference picture P1.
Perform motion compensation prediction as in -8). Block C performs motion compensation prediction with a motion vector of (0, 0) with respect to reference picture P0.

Here, the reference index “0” in the direction 0 indicates the reference picture P0, and the reference index “0” in the direction 1 indicates the reference picture P1.

The reference index refIdxL0X in the direction 0 of the encoding target block X is 0, the motion vector mvL0X is (−8, 8), the reference index refIdxL1X in the direction 1 is 0, and the motion vector mvL1B is (8, −8).

The reference index refIdxL0A in the direction 0 of the peripheral block A is -1, the motion vector mvL0A is (0, 0), the reference index refIdxL1A in the direction 1 is 0, and the motion vector mvL1A is (0, 0).

Reference index refIdxL0B of peripheral block B in direction 0 is 0, motion vector m
vL0B is (−8, 8), the reference index refIdxL1B in the direction 1 is 0, and the motion vector mvL1B is (8, −8).

The reference index refIdxL0C in the direction 0 of the peripheral block C is 0, and the motion vector m
vL0C is (0, 0), the reference index refIdxL1C in the direction 1 is -1, and the motion vector mvL1C is (0, 0).

Here, refIdxL0A is -1, mvL0A is (0, 0), refIdxL1C is -1, and mvL1C is (0, 0) because block A and block C are unidirectionally predicted.

In the conventional motion vector prediction method described above, the direction 0 of the encoding target block X is
Since there are two reference indexes having a value equal to the reference index refIdxL0X of the encoding target block X, the reference index refIdxL0B of the block B and the reference index refIdxL0C of the block C, the prediction motion vector mvpL0X
Is the motion vectors mvL0A, mvL0 in the direction 0 of the peripheral blocks A, B, C as shown in Equation 2.
By calculating the median value of each component of B and mvL0C, (0, 0) is obtained.
mvpL0X = (Media [0, −8,0], Median [0,8,0]) = (
0,0) (Formula 2)

At this time, the difference motion vector dmvL0X is calculated by calculating the difference between the motion vector mvL0X of the encoding target block X and the predicted motion vector mvpL0X as shown in Equation 3.
−8,8).
dmvL0X = mvL0X-mvpL0X = (-8-0,8-0) = (-8,8)
(Formula 3)

Similarly, for the direction 1 of the encoding target block X, the reference index having a value equal to the reference index refIdxL1X of the encoding target block X is 2 of the reference index refIdxL1A of the block A and the reference index refIdxL1B of the block B.
Therefore, the predicted motion vector mvpL1X is represented by the peripheral block A as shown in Equation 4.
By calculating the median value of each component of the motion vectors mvL1A, mvL1B, and mvL1C in the direction 1 of, B, and C, (0, 0) is obtained.
mvpL1X = (Media [0,8,0], Median [0, −8,0]) = (
0,0) (Formula 4)

At this time, by calculating the difference between the motion vector mvL1X of the encoding target block X and the predicted motion vector mvpL1X as shown in Equation 5 by the difference motion vector dmvL1X, (8
-8).
dmvL1X = mvL1X-mvpL1X = (8-0, -8-0) = (8, -8)
(Formula 5)

As described above, the encoding target block X that is bi-directionally predicted has a high correlation with the only peripheral block B that is bi-predicted in the same manner, but the reference index is independently determined for each prediction direction. When the identity is determined, there are peripheral blocks having the same reference index in addition to the peripheral block B. Therefore, in the conventional motion vector prediction method, the predicted motion vector is obtained from the intermediate values of all the neighboring blocks for each prediction direction, and an accurate predicted motion vector cannot be obtained, and the motion vector of the encoding target block Cannot be encoded with reduced redundancy.

The first embodiment of the present invention will be described below in detail based on the above-described prerequisite technology and conventional technology.

FIG. 4 is a diagram illustrating a configuration of a moving image encoding apparatus using the motion vector prediction method according to the first embodiment. The moving image coding apparatus includes a subtraction unit 100, a prediction error coding unit 110, a code string generation unit 120, a prediction error decoding unit 130, an addition unit 140, a motion compensation prediction method determination unit 150, a frame memory 160, a motion vector coding. Unit 170, motion information storage unit 180, and motion vector number acquisition unit 190.

  First, the flow of the video encoding process by the video encoding device will be described.

Encoding is performed in units of macroblocks. The input picture 10 is divided into macroblock units and supplied to the subtraction unit 100 and the motion compensation prediction method determination unit 150.

The motion compensation prediction method determination unit 150 includes the encoding target macroblock and the frame memory 160.
A motion vector is detected for each reference picture from a reference picture supplied by a method such as block matching, and an optimal motion compensation prediction method is determined from the detected motion vectors by a rate distortion optimization method or the like. Is generated.

The motion compensation prediction method determination unit 150 supplies the predicted picture generated from the determined motion compensation prediction method to the subtraction unit 100 and the addition unit 140. Also, the motion compensation prediction method determination unit 150
Sends a reference index and a motion vector corresponding to the determined motion compensation prediction method to the motion information storage unit 180.

Here, the determination of the motion compensation prediction method is a method of dividing a prediction block in a macroblock,
This is to determine a prediction method (unidirectional prediction / bidirectional prediction) of the divided prediction blocks and a reference picture (reference index) in each prediction direction.

The subtraction unit 100 obtains a prediction error based on the difference between the encoding target macroblock and the prediction picture supplied from the motion compensation prediction method determination unit 150, and supplies the prediction error to the prediction error encoding unit 110.

The prediction error encoding unit 110 performs processing such as orthogonal transformation and quantization on the prediction error supplied from the subtraction unit 100 to convert it into prediction error code data, and a code string generation unit 120 and a prediction error decoding unit 130. To supply.

The prediction error decoding unit 130 performs processing such as inverse quantization and inverse orthogonal transformation on the prediction error code data supplied from the prediction error encoding unit 110 to convert the prediction error into prediction error, and the prediction error is sent to the addition unit 140. Supply.

The adding unit 140 obtains a decoded picture by adding the prediction error supplied from the prediction error decoding unit 130 and the prediction picture supplied from the motion compensation prediction method determination unit 150, and supplies the decoded picture to the frame memory 160.

The frame memory 160 holds the decoded picture supplied from the adding unit 140 as a reference picture and supplies it to the motion compensation prediction method determining unit 150.

The motion information storage unit 180 stores the motion vector and the reference index supplied from the motion compensation prediction method determination unit 150, and supplies the motion vector to the motion vector encoding unit 170. The motion information storage unit 180 stores motion vectors and reference indexes for all macroblocks of the input picture, and supplies the motion vectors and reference indexes of neighboring macroblocks of the encoding target macroblock to the motion vector encoding unit 170. To do.

In the first embodiment, the motion information storage unit 180 stores the motion vectors and reference indexes for all macroblocks of the input picture, but the motion information storage unit 1
The purpose of 80 is to supply the motion vectors and reference indexes of the peripheral blocks of the macroblock to be encoded to the motion vector encoding unit 170, and if this can be realized, the motion vectors for all the macroblocks of the input picture. There is no need to store the reference index.

The motion vector encoding unit 170 is supplied from the motion vector and reference index of the encoding target macroblock supplied from the motion information storage unit 180, the motion vector and reference index of the neighboring macroblock, and the motion vector number acquisition unit 190. Determine the motion vector predictor from the number of motion vectors of the encoding target macroblock and neighboring macroblocks, determine the difference motion vector from the motion vector of the encoding target macroblock and the prediction motion vector, convert it to differential motion vector code data, and code string It supplies to the production | generation part 120. A detailed motion vector prediction method will be described later.

The code string generation unit 120 includes prediction error code data supplied from the prediction error encoding unit 110,
The differential motion vector code data supplied from the motion vector encoding unit 170 is converted into an output code string together with other control information and output.

The motion vector number acquisition unit 190 acquires the number of motion vectors for each macro block from the motion information storage unit 180 and supplies the motion vector number of the encoding target macro block and the surrounding macro blocks to the motion vector encoding unit 170.

Regarding the acquisition of the number of motion vectors, in addition to the method of acquiring the number of motion vectors for each macroblock from the motion information storage unit 180, information indicating the motion compensation prediction method determined by the motion compensation prediction method determination unit 150, etc. It is also possible to obtain from.

FIG. 5 is a diagram illustrating a detailed configuration of the motion vector number acquisition unit 190 and the motion vector encoding unit 170 of FIG.

The motion vector number acquisition unit 190 includes an encoding target block motion vector number acquisition unit 210, a peripheral block motion vector number acquisition unit 220, and a peripheral block identification unit 230.

The motion vector encoding unit 170 includes a predicted motion vector deriving unit 240, a differential motion vector calculating unit 250, and a differential motion vector encoding unit 260.

The encoding target block motion vector number acquisition unit 210 acquires the number of motion vectors of the encoding target block from the motion information storage unit 180. The peripheral block specifying unit 230 refers to the motion information storage unit 180 and specifies a plurality of encoded peripheral blocks around the encoding target block. The peripheral block motion vector number acquisition unit 220 acquires the number of motion vectors of each specified peripheral block from the motion information storage unit 180.

The motion vector predictor deriving unit 240 receives information on the reference indexes and motion vectors of a plurality of neighboring blocks from the neighboring block specifying unit 230, and obtains the number of motion vectors of the coding target block from the coding target block motion vector number obtaining unit 210. The number of motion vectors of peripheral blocks is received from the peripheral block motion vector number acquisition unit 220.

The prediction motion vector deriving unit 240 identifies a peripheral block as a prediction candidate block when there is only one peripheral block having the same number of motion vectors as the number of motion vectors of the encoding target block among a plurality of peripheral blocks. Then, the motion vector of the prediction candidate block is supplied to the differential motion vector calculation unit 250 as the predicted motion vector MVP.

The difference motion vector calculation unit 250 calculates a difference motion vector MVD based on the difference between the motion vector MV of the encoding target block read from the motion information storage unit 180 and the prediction motion vector MVP received from the prediction motion vector derivation unit 240. And supplied to the differential motion vector encoding unit 260.

The difference motion vector encoding unit encodes the difference motion vector MVD, and generates a code string generation unit 12.
Supply to zero.

Next, the details of the method for deriving the motion vector predictor according to the first embodiment will be described with reference to FIGS. The prediction motion vector is derived for each prediction block in the encoding target macroblock. In the prediction motion vector derivation method according to the first embodiment, the first prediction motion vector derivation process is performed in the previous stage, and then the second prediction motion vector derivation process is performed in the subsequent stage. The second predictive motion vector derivation process is the same as the conventional predictive motion vector derivation method described in FIG.

  The first prediction motion vector derivation process in the previous stage will be described.

First, the peripheral block specifying unit 230 determines block A, block B, and block C that are peripheral blocks of the encoding target block X as shown in FIG. 1A (step S1).
000).

Next, the peripheral block motion vector number acquisition unit 220 acquires the number of motion vectors of the block A, which is a peripheral block, the number of motion vectors of the block B, and the number of motion vectors of the block C,
The motion vector predictor deriving unit 240 encodes the encoding target block X among the peripheral blocks A, B, and C.
It is checked whether there is only one peripheral block N having the same number of motion vectors as the number of motion vectors (step S1100).

When there is only one peripheral block N having the same number of motion vectors as the encoding target block X (YES in step S1100), the motion vector predictor deriving unit 240 identifies the peripheral block as a prediction candidate block, and It is checked whether or not the number of motion vectors in block X is 2 (step S1200).

When only one peripheral block N has the same number of motion vectors as the encoding target block X and the number of motion vectors of the encoding target block X is 2 (YES in step S1200),
As in Expression 6, the motion vector of the peripheral block N is set as the predicted motion vector of the encoding target block X in both directions (forward L0 and backward L1) (step S1300).
mvpL0X = mvL0N
mvpL1X = mvL1N (Formula 6)

When only one peripheral block N has the same number of motion vectors as the encoding target block X and the number of motion vectors of the encoding target block X is not 2 (NO in step 1200), Expression 7
As described above, the motion vector in the single prediction direction Z (forward L0 or backward L1) of the block N is set as the predicted motion vector of the encoding target block X (step S1400). The single prediction direction Z is a direction in which a motion vector exists.
mvpLZX = mvLZN (Formula 7)

When none of the peripheral blocks has the same number of motion vectors as the encoding target block X, or when two or more peripheral blocks have the same number of motion vectors as the encoding target block X (S10)
00, NO), as the second predicted motion vector derivation process, the conventional process after step 5000 is performed.

In the above description, the peripheral blocks, block A, block B, and block C, are
Although it was in the position shown in FIG. 1 (A), the peripheral block is not limited to the position shown in FIG. 1 (A), and may be in any position as long as it is an encoded block. Further, the number of peripheral blocks is not necessarily three, but may be two or more. Furthermore, the peripheral block is not only a block in the same picture that is spatially adjacent to the current block, but also a reference picture that is encoded immediately before it is temporally adjacent to the current block. It may be a block at the same position as the block or a block adjacent thereto.

In the first embodiment, as the second predicted motion vector derivation process, the process after step 5000 is performed. However, the process is not limited to the process after step 5000. For example, when two or more peripheral blocks have the same number of motion vectors as the encoding target block X, the motion vector of any one of a plurality of peripheral blocks having the same number of motion vectors as the encoding target block X May be used as a predicted motion vector, or a weighted average of motion vectors of a plurality of peripheral blocks having the same number of motion vectors as the encoding target block X may be obtained and used as a predicted motion vector. Similarly, even when none of the neighboring blocks has the same number of motion vectors as the encoding target block X, the motion vector of any one of the neighboring blocks may be used as a predicted motion vector, and the motion vectors of a plurality of neighboring blocks may be used. A weighted average may be obtained and used as a predicted motion vector.

  Next, the effect by the prediction motion vector deriving method of the first embodiment will be described.

It is assumed that the prediction motion vector derivation method of the first embodiment is applied to the specific example described in FIG. Since the only peripheral block having the same number of motion vectors as the number of motion vectors 2 of the encoding target block X is the block B, the predicted motion vector mv of the encoding target block X
pL0X and mvpL1X are (−8, 8) and (8, −8) from Equation 8, respectively.
mvpL0X = mvL0B = (− 8,8)
mvpL1X = mvL1B = (8, −8) (Formula 8)

At this time, the difference motion vectors dmvL0X and dmvL1X are expressed by (0, 0
), (0, 0).
dmvpL0X = mvL0X-mvpL0X = (− 8,8) − (− 8,8) = (0,0)
)
dmvpL1X = mvL0X−mvpL0X = (8, −8) − (8, −8) = (0,0)
(Formula 9)

As described above, according to the prediction motion vector derivation method of the first embodiment, the motion vector of the peripheral block having the same number of motion vectors as the number of motion vectors of the encoding target block is used as the prediction motion vector. Even if the motion of an object having a temporal correlation different from the motion of the encoding target block exists in the neighboring blocks, the motion vector redundancy of the encoding target block is derived by deriving the predicted motion vector according to the original motion. Can be encoded.

FIG. 7 is a diagram illustrating a configuration of a moving picture decoding apparatus using the motion vector prediction method according to the first embodiment. The video decoding apparatus includes a code string analysis unit 500, a prediction error decoding unit 510, and an addition unit 5.
20, a motion compensation prediction unit 530, a frame memory 540, a motion vector decoding unit 550, a motion information storage unit 560, and a motion vector number acquisition unit 570.

  First, the flow of moving picture decoding processing by the moving picture decoding apparatus will be described.

Decoding is performed in units of macroblocks. The input code string 10 is supplied to the code string analysis unit 500.

The code string analysis unit 500 decodes and obtains prediction error encoded data, a difference motion vector, a reference index, and a macroblock type from the input code string 10. Code string analysis unit 5
00 supplies the prediction error encoded data to the prediction error decoding unit 510, the difference motion vector and the reference index to the motion vector decoding unit 550, and the macroblock type to the motion vector number acquisition unit 570 in units of prediction blocks.

For example, when the macro block is composed of four prediction blocks, the subsequent processing is repeated four times within the macro block.

The prediction error decoding unit 510 converts the prediction error encoded data supplied from the code string analysis unit 500 into a prediction error by performing processing such as inverse quantization and inverse orthogonal transform, and supplies the prediction error to the addition unit 520. .

The motion vector decoding unit 550 includes the difference motion vector and reference index of the decoding target macroblock supplied from the code string analysis unit 500, and the number of motion vectors of the decoding target macroblock and neighboring macroblocks supplied from the motion vector number acquisition unit 570. , Motion information storage unit 56
A prediction motion vector is obtained using the motion vector and reference index of the neighboring macroblock supplied from 0, the motion vector is decoded from the difference motion vector and the prediction motion vector of the decoding target macroblock, and the decoded motion vector and reference index are obtained. This is supplied to the motion compensation prediction unit 530 and also supplied to the motion information storage unit 560. A detailed motion vector decoding method will be described later.

The motion information storage unit 560 stores the motion vector and reference index supplied from the motion vector decoding unit 550, and supplies the stored motion vector and reference index to the motion vector decoding unit 550 as necessary.

The motion vector number acquisition unit 570 acquires the number of motion vectors for each macroblock from the macroblock type supplied from the code stream analysis unit 500, and calculates the number of motion vectors of the decoding target macroblock and the surrounding macroblocks as the motion vector decoding unit 550. To supply.

The acquisition of the number of motion vectors can be determined from the number of difference motion vectors and the number of reference indexes in addition to the acquisition from the macroblock type.

The motion compensated prediction unit 530 generates a prediction value using the motion vector and reference index supplied from the motion vector decoding unit 550 and the reference picture supplied from the frame memory 540, and supplies the prediction value to the addition unit 520.

The addition unit 520 is provided with the prediction error supplied from the prediction error decoding unit 510 and the motion compensation prediction unit 53.
A predicted value supplied from 0 is added to generate a decoded value, which is supplied to the frame memory 540 and output as an image output 60.

The frame memory 540 stores the decoded picture supplied from the adding unit 520, and supplies the stored decoded picture as a reference picture to the motion compensation prediction unit 530 as necessary.

FIG. 8 is a diagram illustrating a detailed configuration of the motion vector number acquisition unit 570 and the motion vector decoding unit 550 of FIG.

The motion vector number acquisition unit 570 includes a decoding target block motion vector number acquisition unit 610, a peripheral block motion vector number acquisition unit 620, and a peripheral block identification unit 630.

The motion vector decoding unit 550 includes a predicted motion vector derivation unit 640 and a motion vector calculation unit 650.

The decoding target block motion vector number acquisition unit 610 acquires the number of motion vectors of the decoding target block from the code string analysis unit 500. The peripheral block specifying unit 230 includes the motion information storage unit 5
Referring to 60, a plurality of peripheral blocks that have been decoded and are in the vicinity of the decoding target block are identified. The peripheral block motion vector number acquisition unit 620 acquires the number of motion vectors of each specified peripheral block from the motion information storage unit 560.

The motion vector predictor deriving unit 640 receives information on reference indexes and motion vectors of a plurality of neighboring blocks from the neighboring block specifying unit 630, and obtains the number of motion vectors of the decoding target block from the decoding target block motion vector number obtaining unit 610. The number of motion vectors of surrounding blocks is received from the motion vector number acquisition unit 620.

The predicted motion vector deriving unit 640 specifies, as a prediction candidate block, when there is only one peripheral block having the same number of motion vectors as the number of motion vectors of the decoding target block among the plurality of peripheral blocks. The motion vector of the prediction candidate block is supplied to the motion vector calculation unit 650 as a predicted motion vector MVP.

The motion vector calculation unit 650 calculates the motion vector MV by adding the difference motion vector MVD of the decoding target block received from the code string analysis unit 500 and the prediction motion vector MVP received from the prediction motion vector derivation unit 640. And supplied to the motion compensation prediction unit 530.

The method of deriving a predicted motion vector by the moving picture decoding apparatus according to the first embodiment is based on the moving picture coding apparatus according to the first embodiment if “the encoding target block” is read as “the decoding target block”. Since it is the same as the method of deriving the predicted motion vector, detailed description is omitted.

  Next, a second embodiment of the present invention will be described.

FIG. 9 is a diagram illustrating a configuration of a moving image encoding apparatus using the motion vector prediction method according to the second embodiment. The moving image coding apparatus includes a subtraction unit 100, a prediction error coding unit 110, a code string generation unit 120, a prediction error decoding unit 130, an addition unit 140, a motion compensation prediction method determination unit 150, a frame memory 160, a motion vector coding. Unit 170, motion information storage unit 180, motion vector number acquisition unit 190, and macroblock boundary determination unit 200.

Subtraction unit 100, prediction error encoding unit 110, code sequence generation unit 120, prediction error decoding unit 130, addition unit 140, motion compensation prediction method determination unit 150, frame memory 1 of the second exemplary embodiment
60, the motion information storage unit 180, and the motion vector number acquisition unit 190 have the same configurations as those in the first embodiment. A motion vector encoding unit 170 and a macroblock boundary determination unit 200 different from those in the first embodiment will be described with reference to FIGS. 9 and 10.

The macroblock boundary determining unit 200 determines whether or not the encoding target block touches the macroblock boundary. If the encoding target block touches the macroblock boundary, the macroblock boundary flag is set to 1, and the encoding target block is a macro. When the block boundary is not touched, the macroblock boundary flag is set to 0 and the macroblock boundary flag is supplied to the motion vector encoding unit 170.

Whether or not to contact the macroblock boundary can be determined using, for example, a prediction block number. The prediction block number will be described later.

The motion vector encoding unit 170 is supplied from the motion information storage unit 180, the motion vector and reference index of the encoding target macroblock, the motion vector and reference index of the surrounding macroblock, and the motion supplied from the motion vector number acquisition unit 190. A prediction motion vector is obtained from the number of vectors and the macroblock boundary flag supplied from the macroblock boundary determination unit 200, and a difference motion vector is obtained from the motion vector of the encoding target macroblock and the prediction motion vector to obtain difference motion vector code data. The data is converted and supplied to the code string generation unit 120.

A detailed motion vector prediction method according to the second embodiment will be described with reference to FIGS.

First, the peripheral block specifying unit 230 determines block A, block B, block C, and block D, which are peripheral blocks of the encoding target block X, as shown in FIG. 1B (step S2000).

Next, the macroblock boundary determination unit 200 checks the value of the macroblock boundary flag (step S2100).

If the macroblock boundary flag is 1 (YES in step S2100), the peripheral block motion vector number acquisition unit 220 acquires the number of motion vectors of the block A, block B, block C, and block D, which are peripheral blocks, The predicted motion vector deriving unit 240
It is checked whether or not there is only one peripheral block N having the same number of motion vectors as that of the encoding target block X among the peripheral blocks A, B, C, and D (step S).
2200). After step S2200, the processing after step S1200 of the first embodiment is performed.

If the macroblock boundary flag is 0 (NO in step S2100), as the second motion vector predictor deriving process, the processes in and after step S5000 are performed as described in the first embodiment.

  The effect of the motion vector prediction method of the second embodiment will be described with reference to FIG.

First, the prediction block number will be described. In FIG. 11A, the inside of the macroblock is 8x
8 shows a prediction block number when divided into four prediction blocks.

As shown in FIG. 11B, there is a block E having a prediction block number of 3 that does not touch the macroblock boundary. When the block E includes the same object as the other blocks in the macro block and exhibits the same movement, the block E has one block X as shown in FIG. 11C, FIG. 11D, or FIG. Are likely to be encoded. Therefore, when the block E exists as an independent prediction block, it is unlikely that the block E and other blocks in the macro block include the same object and exhibit the same motion. So in this case,
It is desirable to proceed to the second predictive motion vector derivation process without making a determination based on the number of motion vectors, so that sudden motion vectors can be eliminated.

For a prediction block that does not touch the macroblock boundary, such as block E, since the neighboring block exists in the same macroblock, the condition that it does not touch the macroblock boundary is that “the peripheral block is in the same macroblock. In other words.

According to the prediction motion vector derivation method of this embodiment, when the encoding target block is in contact with the macroblock boundary, the encoding target block is considered in consideration of the number of motion vectors of the encoding target block and the number of motion vectors of the surrounding blocks. When the block does not touch the macroblock boundary, the number of motion vectors of the current block and the number of motion vectors of the neighboring blocks are not taken into consideration, thereby reducing false detection of predicted motion vectors and predicting motion vectors according to the original motion. Can be derived.

FIG. 12 is a diagram illustrating a configuration of a moving image decoding apparatus using the motion vector prediction method according to the second embodiment. The moving picture decoding apparatus includes a code string analysis unit 500, a prediction error decoding unit 510, an addition unit 520, a motion compensation prediction unit 530, a frame memory 540, a motion vector decoding unit 550, a motion information storage unit 560, and a motion vector number acquisition unit 570. , And macroblock boundary determination unit 5
80.

The prediction error decoding unit 510, the addition unit 520, the motion compensation prediction unit 530, the frame memory 540, the motion information storage unit 560, and the motion vector number acquisition unit 570 of the second embodiment are the same as those of the first embodiment. It is the same as the configuration marked with. A code string analysis unit 500, a motion vector decoding unit 550, and a macroblock boundary determination unit 580 that are different from those in the first embodiment will be described.

The code string analysis unit 500 decodes and obtains prediction error encoded data, a difference motion vector, a reference index, and a macroblock type from the input code string 10. Code string analysis unit 5
00 is the prediction block unit, the prediction error encoded data in the prediction error decoding unit 510, the difference motion vector and the reference index in the motion vector decoding unit 550, the macroblock type in the motion vector number acquisition unit 570, and the prediction block number. Is supplied to the macroblock boundary determination unit 580.

The macroblock boundary determination unit 580 determines whether or not the decoding target block touches the macroblock boundary from the prediction block number supplied from the code string analysis unit 500, and if the decoding target block touches the macroblock boundary, the macroblock boundary If the flag is set to 1 and the decoding target block does not touch the macroblock boundary, the macroblock boundary flag is set to 0 and the macroblock boundary flag is supplied to the motion vector decoding unit 550.

The motion vector decoding unit 550 includes the difference motion vector and reference index of the decoding target macroblock supplied from the code string analysis unit 500, and the number of motion vectors of the decoding target macroblock and neighboring macroblocks supplied from the motion vector number acquisition unit 570. Then, a prediction motion vector is obtained using the macroblock boundary flag supplied from the macroblock boundary determination unit 580, the motion vector of the neighboring blocks supplied from the motion information storage unit 560, and the reference index, and the difference motion vector of the decoding target macroblock The motion vector is decoded from the predicted motion vector, and the decoded motion vector and the reference index are supplied to the motion compensation prediction unit 530 and supplied to the motion information storage unit 560.

The method of deriving a predicted motion vector by the moving picture decoding apparatus according to the second embodiment is based on the moving picture coding apparatus according to the second embodiment, if “encoding target block” is read as “decoding target block”. Since it is the same as the method of deriving the predicted motion vector, detailed description is omitted.

Next, before describing the third embodiment of the present invention, the problem of the conventional motion vector prediction method will be described with another specific example with reference to FIG.

The positional relationship of the surrounding block A, block B, and block C with respect to the encoding target block X is as illustrated. In this example, two forward reference pictures P0 and P1 and one backward reference picture P2 are used for the current picture. Since the same object exists in the encoding target block X and the block B, bi-directional prediction using the forward reference picture P1 and the backward reference picture P2 is used for the encoding target block X and the block B. On the other hand, the object including the block A includes a forward reference picture P0 and a backward reference picture P.
2 bidirectional prediction is used, and the object including the block C uses the unidirectional prediction of the forward reference picture P1.

The encoding target block X performs motion compensation prediction with a motion vector (−8, 8) for the reference picture P1, and motion compensation prediction with a motion vector (8, −8) for the reference picture P2. I do. The block A has a motion vector (0, 0) with respect to the reference picture P0.
The motion compensation prediction is performed so that the motion vector is (0, 0) with respect to the reference picture P2. The block B has a motion vector (−8,
8), and the motion vector is (8, −8) for the reference picture P2.
Motion compensation prediction is performed. Block C has a motion vector of (
0,0) is performed.

Here, the reference index “0” in the direction 0 indicates the reference picture P1, the reference index “1” in the direction 0 indicates the reference picture P0, and the reference index “0” in the direction 1 indicates the reference picture P2. And

The reference index refIdxL0X in the direction 0 of the encoding target block X is 0, the motion vector mvL0X is (−8, 8), the reference index refIdxL1X in the direction 1 is 0, and the motion vector mvL1B is (8, −8).

The reference index refIdxL0A in the direction 0 of the peripheral block A is 1, the motion vector motion vector mvL0A is (0, 0), and the reference index refIdxL1A in the direction 1 is 0.
, MvL1A is (0, 0).

Reference index refIdxL0B of peripheral block B in direction 0 is 0, motion vector m
vL0B is (−8, 8), the reference index refIdxL1B in the direction 1 is 0, and the motion vector mvL1B is (8, −8).

The reference index refIdxL0C in the direction 0 of the peripheral block C is 0, and the motion vector m
vL0C is (0, 0), the reference index refIdxL1C in the direction 1 is -1, and the motion vector mvL1C is (0, 0).

In the conventional motion vector prediction method, for the direction 0 of the encoding target block X, the reference index having a value equal to the reference index refIdxL0X of the encoding target block X is the reference index refIdxL0B of the block B and the reference index refIdxL0C of the block C. Therefore, the predicted motion vector mvpL0X is represented by the motion vectors mvL0A, mvL0B in the direction 0 of the peripheral blocks A, B, and C as shown in Equation 10.
By calculating the median value of each component of mvL0C, (0, 0) is obtained.
mvpL0X = (Media [0, −8,0], Median [0,8,0]) = (
0,0) (Equation 10)

At this time, the difference motion vector dmvL0X becomes (−8, 8) by calculating the difference between the motion vector mvL0X of the encoding target block X and the predicted motion vector mvpL0X as shown in Equation 11.
dmvL0X = mvL0X-mvpL0X = (-8-0,8-0) = (-8,8)
(Formula 11)

Similarly, for the direction 1 of the encoding target block X, the reference index having a value equal to the reference index refIdxL1X of the encoding target block X is 2 of the reference index refIdxL1A of the block A and the reference index refIdxL1B of the block B.
Therefore, the predicted motion vector mvpL1X is obtained by calculating the median value of each component of the motion vectors mvL1A, mvL1B, and mvL1C in the direction 1 of the peripheral blocks A, B, and C, as shown in Equation 12, (0, 0).
mvpL1X = (Media [0,8,0], Median [0, −8,0]) = (
0,0) (Formula 12)

At this time, by calculating the difference between the motion vector mvL1X of the encoding target block X and the predicted motion vector mvpL1X as shown in Equation 12, the difference motion vector dmvL1X (
8, -8).
dmvL1X = mvL1X-mvpL1X = (8-0, -8-0) = (8, -8)
(Formula 13)

As described above, the encoding target block X that is bi-directionally predicted has a high correlation with the only peripheral block B that is bi-predicted in the same manner, but the reference index is independently determined for each prediction direction. When the identity is determined, there are peripheral blocks having the same reference index in addition to the peripheral block B. Therefore, in the conventional motion vector prediction method, the predicted motion vector is obtained from the intermediate values of all the neighboring blocks for each prediction direction, and an accurate predicted motion vector cannot be obtained, and the motion vector of the encoding target block Cannot be encoded with reduced redundancy.

Further, in this case, even if the same number of motion vectors is estimated for the encoding target block and the surrounding blocks using the motion vector prediction method according to the first or second embodiment of the present invention, the encoding target block Since there are two blocks A and B as peripheral blocks equal to the number of motion vectors 2, the second motion vector prediction process is executed, and the same result as the conventional motion vector prediction method is obtained. Coding with reduced redundancy of motion vector of the target block is not possible.

In order to cope with such a further problem, in the third embodiment of the present invention, not only the number of motion vectors is the same between the current block and the neighboring blocks, but also the reference picture is the same. The motion vector prediction method put into the evaluation is used.

  Hereinafter, the third embodiment of the present invention will be described in detail.

The configuration of the moving image encoding device and the moving image decoding device using the motion vector prediction method of the third embodiment is the same as that of the moving image encoding device and the moving image decoding device of the first embodiment.
A motion vector encoding unit 170 and a motion vector decoding unit 550 having different operations from those of the first embodiment will be described with reference to FIG. However, in the case of the motion vector decoding unit 550, “encoding target block” is replaced with “decoding target block” in the following description.

  The processing after step S1200, which is different from the first embodiment, will be described.

When there is only one peripheral block N having the same number of motion vectors as the encoding target block X, the motion vector predictor deriving unit 240 checks whether the number of motion vectors of the encoding target block X is 2 ( Step S1200).

When the number of motion vectors of the encoding target block X is 2 (YES in step S1200)
), The motion vector predictor deriving unit 240 determines that the reference index refIdxL0X in the L0 direction of the encoding target block X is the reference index refIdxL in the L0 direction of the neighboring block N.
Reference index refIdxL in the L1 direction of the encoding target block X equal to 0N
It is checked whether 1X is equal to the reference index refIdxL1N in the L1 direction of the peripheral block N (step S3000).

If the condition of step S3000 is satisfied, step S3000 is performed as in the first embodiment.
The processing of 1300 is performed, and the motion vector of the peripheral block N is set as the predicted motion vector of the encoding target block X in both directions (forward L0 and backward L1). When the condition of step S3000 is not satisfied, the processing after step S5000, which is a conventional motion vector prediction method, is performed.

When the number of motion vectors of the encoding target block X is not 2 (NO in step 1200),
The motion vector predictor deriving unit 240 checks whether or not the reference index refIdxLZX in the single prediction direction Z (forward L0 or backward L1) of the encoding target block X is equal to the reference index refIdxLZN in the LZ direction of the neighboring block N (step). S3100)
.

If the condition of step S3100 is satisfied, step S3 is performed as in the first embodiment.
The processing of 1400 is performed, and the motion vector in the single prediction direction Z of the block N is set as the prediction motion vector of the encoding target block X. If the condition of step S3100 is not satisfied, the process after step S5000, which is a conventional motion vector prediction method, is performed.

  The effect by the prediction motion vector deriving method of the third embodiment will be described.

It is assumed that the prediction motion vector deriving method of the third embodiment is applied to the specific example described in FIG. Since there are two blocks A and B as peripheral blocks having the same number of motion vectors as the number of motion vectors 2 of the encoding target block X, the only peripheral block having the same reference index in both directions is the block B. Predicted motion vectors mvpL0X and mvpL1X of the encoding target block X are (−8, 8) and (8, −8) from Expression 14, respectively.
mvpL0X = mvL0B = (− 8,8)
mvpL1X = mvL1B = (8, −8) (Formula 14)

At this time, the difference motion vectors dmvL0X and dmvL1X are expressed by (0,
0), (0, 0).
dmvpL0X = mvL0X-mvpL0X = (− 8,8) − (− 8,8) = (0,0)
)
dmvpL1X = mvL0X−mvpL0X = (8, −8) − (8, −8) = (0,0)
(Formula 15)

As described above, according to the prediction motion vector derivation method of the third embodiment, even when there are a plurality of peripheral blocks having the same number of motion vectors as the number of motion vectors of the encoding target block, reference to each prediction direction is performed. By considering whether the index is the same, that is, whether the reference picture of the encoding target block and the surrounding block match, a prediction motion vector according to the original motion is derived, and the motion vector of the encoding target block Can be encoded with reduced redundancy.

In any of the embodiments, the number and positions of the peripheral blocks are arbitrary. Blu-ray (trademark)
In a general recording medium such as a disk, the front L0 reference pictures are many, and the rear L1 reference pictures are few. In such a case, the number of neighboring blocks may be changed depending on the prediction direction, for example, 4 blocks are specified for the forward L0 and 2 blocks are specified for the backward L1.

Also, in any method of deriving a motion vector predictor according to any embodiment, a first motion vector predictor derivation process in the previous stage, a second motion vector predictor derivation process in the subsequent stage (conventional motion vector derivation method), By changing the peripheral block specifying method, the peripheral block specifying method can be adaptively selected according to the derivation accuracy and processing amount of the predicted motion vector.

The moving image encoded stream output from the moving image encoding apparatus of the embodiment described above has a specific data format so that it can be decoded according to the encoding method used in the embodiment. Therefore, the moving picture decoding apparatus corresponding to the moving picture encoding apparatus can decode the encoded stream of this specific data format.

When a wired or wireless network is used to exchange an encoded stream between a moving image encoding device and a moving image decoding device, the encoded stream is converted into a data format suitable for the transmission form of the communication path. It may be transmitted. In that case, a moving image transmitting apparatus that converts the encoded stream output from the moving image encoding apparatus into encoded data in a data format suitable for the transmission form of the communication channel and transmits the encoded data to the network, and receives the encoded data from the network Then, a moving image receiving apparatus that restores the encoded stream and supplies the encoded stream to the moving image decoding apparatus is provided.

The moving image transmitting apparatus is a memory that buffers the encoded stream output from the moving image encoding apparatus, a packet processing unit that packetizes the encoded stream, and transmission that transmits the packetized encoded data via the network. Part. The moving image receiving apparatus generates a coded stream by packetizing the received data, a receiving unit that receives the packetized coded data via a network, a memory that buffers the received coded data, and packet processing. And a packet processing unit provided to the video decoding device.

The above processing relating to encoding and decoding can be realized as a transmission, storage, and reception device using hardware, and is stored in a ROM (Read Only Memory), a flash memory, or the like. It can also be realized by firmware or software such as a computer. The firmware program and software program can be recorded on a computer-readable recording medium, provided from a server through a wired or wireless network, or provided as a data broadcast of terrestrial or satellite digital broadcasting Is also possible.

The present invention has been described based on the embodiments. The embodiments are exemplifications, and it will be understood by those skilled in the art that various modifications can be made to combinations of the respective constituent elements and processing processes, and such modifications are within the scope of the present invention. .

100 subtraction unit, 110 prediction error encoding unit, 120 code string generation unit, 130
Prediction error decoding unit, 140 addition unit, 150 compensated prediction method determination unit, 160 frame memory, 170 motion vector encoding unit, 180 motion information storage unit, 190 motion vector number acquisition unit, 200 macroblock boundary determination unit, 210 encoding Target block motion vector number acquisition unit, 220 peripheral block motion vector number acquisition unit, 230
Peripheral block specifying unit, 240 predicted motion vector deriving unit, 250 differential motion vector calculating unit, 260 differential motion vector encoding unit, 500 code string analysis unit, 510 prediction error decoding unit, 520 addition unit, 530 motion compensation prediction unit, 540 Frame memory, 550 motion vector decoding unit, 560 motion information storage unit, 570 motion vector number acquisition unit, 580 macroblock boundary determination unit, 610 decoding target block motion vector number acquisition unit, 620 peripheral block motion vector number acquisition unit, 630 periphery A block specifying unit; 640 a predicted motion vector deriving unit; and 650 a motion vector calculating unit.

Claims (6)

A moving picture decoding apparatus for decoding a code string encoded using motion compensated prediction in units of blocks obtained by dividing each picture of a moving picture,
A motion information deriving unit that derives a motion vector of a decoding target prediction block based on a motion vector of any candidate block among candidate blocks selected from neighboring blocks of the decoding target prediction block;
A motion compensation prediction unit that generates a prediction signal of the decoding target prediction block by motion compensation prediction using a motion vector derived by the motion information derivation unit;
The motion information deriving unit, when the number of motion vectors of the first block and the number of motion vectors of the second block is 2, a reference index indicating a reference picture referred to by the first motion vector of the first block And a reference index indicating a reference picture referenced by a second motion vector of the first block, and a reference index indicating a reference picture referenced by a second motion vector of the first block, and the first index It is determined whether or not the second block is a candidate block depending on whether or not the reference index indicating the reference picture referred to by the second motion vector of the second block is the same,
The moving picture decoding apparatus, wherein the peripheral block includes at least a block adjacent to the upper left of the decoding target prediction block. A moving picture decoding method for decoding a code string encoded using motion compensated prediction in units of blocks obtained by dividing each picture of a moving picture,
A motion information deriving step for deriving a motion vector of a decoding target prediction block based on a motion vector of any one of candidate blocks selected from neighboring blocks of the decoding target prediction block;
A motion compensation prediction step of generating a prediction signal of the decoding target prediction block by motion compensation prediction using a motion vector derived by the motion information deriving unit;
In the motion information deriving step, when the number of motion vectors of the first block and the number of motion vectors of the second block are 2, a reference index indicating a reference picture to which the first motion vector of the first block refers And a reference index indicating a reference picture referenced by a second motion vector of the first block, and a reference index indicating a reference picture referenced by a second motion vector of the first block, and the first index It is determined whether or not the second block is a candidate block depending on whether or not the reference index indicating the reference picture referred to by the second motion vector of the second block is the same,
The video decoding method, wherein the peripheral block includes at least a block adjacent to the upper left of the decoding target prediction block. A moving image decoding program for decoding a code string encoded using motion compensated prediction in units of blocks obtained by dividing each picture of a moving image,
A motion information deriving step for deriving a motion vector of a decoding target prediction block based on a motion vector of any one of candidate blocks selected from neighboring blocks of the decoding target prediction block;
Causing the computer to perform a motion compensation prediction step of generating a prediction signal of the decoding target prediction block by motion compensation prediction using a motion vector derived by the motion information deriving unit;
In the motion information deriving step, when the number of motion vectors of the first block and the number of motion vectors of the second block are 2, a reference index indicating a reference picture to which the first motion vector of the first block refers And a reference index indicating a reference picture referenced by a second motion vector of the first block, and a reference index indicating a reference picture referenced by a second motion vector of the first block, and the first index It is determined whether or not the second block is a candidate block depending on whether or not the reference index indicating the reference picture referred to by the second motion vector of the second block is the same,
The peripheral block includes at least a block adjacent on the upper left of the decoding target prediction block. A receiving device that decodes a code string encoded using motion compensated prediction in units of blocks obtained by dividing each picture of a moving image,
A receiving unit for receiving packetized encoded data via a network;
A memory for buffering received encoded data;
A packet processing unit that packet-processes the encoded data to generate an encoded stream;
A motion information deriving unit that derives a motion vector of a decoding target prediction block in the encoded stream based on a motion vector of any candidate block selected from neighboring blocks of the decoding target prediction block;
A motion compensation prediction unit that generates a prediction signal of the decoding target prediction block by motion compensation prediction using a motion vector derived by the motion information derivation unit;
The motion information deriving unit, when the number of motion vectors of the first block and the number of motion vectors of the second block is 2, a reference index indicating a reference picture referred to by the first motion vector of the first block And a reference index indicating a reference picture referenced by a second motion vector of the first block, and a reference index indicating a reference picture referenced by a second motion vector of the first block, and the first index It is determined whether or not the second block is a candidate block depending on whether or not the reference index indicating the reference picture referred to by the second motion vector of the second block is the same,
The reception apparatus, wherein the peripheral block includes at least a block adjacent to the upper left of the decoding target prediction block. A reception method for decoding a code string encoded using motion compensated prediction in units of blocks obtained by dividing each picture of a moving image,
A reception step of receiving packetized encoded data via a network;
Buffering received encoded data; and
A packet processing step of packet-processing the encoded data to generate an encoded stream;
A motion information deriving step for deriving a motion vector of a decoding target prediction block in the encoded stream based on a motion vector of any candidate block selected from neighboring blocks of the decoding target prediction block;
A motion compensation prediction step of generating a prediction signal of the decoding target prediction block by motion compensation prediction using a motion vector derived by the motion information deriving unit;
In the motion information deriving step, when the number of motion vectors of the first block and the number of motion vectors of the second block are 2, a reference index indicating a reference picture to which the first motion vector of the first block refers And a reference index indicating a reference picture referenced by a second motion vector of the first block, and a reference index indicating a reference picture referenced by a second motion vector of the first block, and the first index It is determined whether or not the second block is a candidate block depending on whether or not the reference index indicating the reference picture referred to by the second motion vector of the second block is the same,
The reception method, wherein the peripheral block includes at least a block adjacent to the upper left of the decoding target prediction block. A reception program for decoding a code string encoded using motion compensated prediction in units of blocks obtained by dividing each picture of a moving image,
A reception step of receiving packetized encoded data via a network;
Buffering received encoded data; and
A packet processing step of packet-processing the encoded data to generate an encoded stream;
A motion information deriving step for deriving a motion vector of a decoding target prediction block in the encoded stream based on a motion vector of any candidate block selected from neighboring blocks of the decoding target prediction block;
Causing the computer to perform a motion compensation prediction step of generating a prediction signal of the decoding target prediction block by motion compensation prediction using a motion vector derived by the motion information deriving unit;
In the motion information deriving step, when the number of motion vectors of the first block and the number of motion vectors of the second block are 2, a reference index indicating a reference picture to which the first motion vector of the first block refers And a reference index indicating a reference picture referenced by a second motion vector of the first block, and a reference index indicating a reference picture referenced by a second motion vector of the first block, and the first index It is determined whether or not the second block is a candidate block depending on whether or not the reference index indicating the reference picture referred to by the second motion vector of the second block is the same,
The reception program according to claim 1, wherein the peripheral block includes at least a block adjacent to the upper left of the decoding target prediction block.

Images