WO2013077659A1

WO2013077659A1 - Method and apparatus for predictive encoding/decoding of motion vector

Info

Publication number: WO2013077659A1
Application number: PCT/KR2012/009964
Authority: WO
Inventors: 임정연; 박중건; 문주희; 이영렬; 김해광; 전병우; 한종기; 임성원; 이주옥
Original assignee: 에스케이텔레콤 주식회사
Priority date: 2011-11-24
Filing date: 2012-11-23
Publication date: 2013-05-30

Abstract

Provided is a method and an apparatus for image encoding/decoding, one embodiment of the present invention comprising: an image encoder which selects at least one representative block from among neighboring blocks, which have been encoded via inter-prediction, on the basis of the order of priority of the inter-predictive method, inter-predictive direction, motion compensation direction, size of the block and the distance from the reference picture, encodes the motion vector of the current block using the motion vector selected as a motion vector predictor from among the motion vectors of the selected representative block and temporal candidate motion vector, and outputs a bit stream; and an image decoder which extracts an index for a motion vector predictor and a residual vector from the bit stream, selects at least one representative block from among the neighboring blocks, which have been encoded via inter-prediction, on the basis of the order of priority of the inter-predictive method, inter-predictive direction, motion compensation direction, size of the block and the distance from the reference picture, identifies the index by considering the selected representative block, decodes the motion vector of the representative block corresponding to the index as the motion vector predictor of the current block, and decodes the current block.

Description

Method and apparatus for predictive coding / decoding of motion vectors

The present invention relates to a method and apparatus for predictive encoding / decoding of motion vectors. More specifically, the present invention relates to a method and apparatus for improving the coding efficiency by predicting the motion vector of the current block more accurately using the motion vectors of neighboring blocks of the current block.

The contents described in this section merely provide background information on the embodiments of the present invention and do not constitute a prior art.

In general, a moving picture is composed of a series of pictures, and each picture is divided into a predetermined area such as a block. When a region of an image is divided into blocks, the divided blocks are largely classified into intra blocks and inter blocks according to encoding methods. An intra block refers to a block that is encoded by using an intra prediction coding scheme. An intra prediction coding is performed by using pixels of blocks that are previously encoded, decoded, and reconstructed in a current picture that performs current encoding. By predicting the pixels of the block, a prediction block is generated and a difference value with the pixels of the current block is encoded. An inter block refers to a block that is encoded using inter prediction coding. Inter prediction coding generates a prediction block by predicting a current block in a current picture by referring to one or more past or future pictures, and then generates a current block. This is a method of encoding the difference value with. Here, a picture referred to for encoding or decoding the current picture is referred to as a reference picture.

In general, the reason for encoding a video signal is to reduce the size (capacity) of the video, that is, to compress the video. In general, image compression is performed by two methods of intra prediction and inter prediction. The purpose of the prediction is to predict a similar signal from the original signal and transmit the difference value. Assuming that there are 100, 200, 150, 240, and 178 pixel values, the binary representation is 100 = 1100100 (2), 200 = 11001000 (2), 150 = 10010110 (2), 240 = 11110000 (2), 178 = 10110010 (2) and a total of 40 bits are used. However, if these values are predicted (assuming that the predicted values are 95, 190, 149, 240, 178) and only the residual signal that is the difference from the original signal is encoded, 5 = 101 (2), 10 = 1010 (2), 1 = 1 (2) and 0 = 0 (2), and thus the number of bits used for encoding is reduced and the size of the entire image transmitted to the decoding apparatus is reduced. In other words, the better the prediction, the higher the coding efficiency.

There are two types of predictors: intra prediction and inter prediction.

Intra prediction is a method of generating a prediction block (prediction signals) using neighboring pixels neighboring a current block (original signal) and then transmitting the residual data (residual signal) which is the difference. In this case, the residual signal and the prediction mode index indicating which direction the prediction is performed may be transmitted (for example, when the prediction is performed using the upper block, the prediction mode is 0, and when the prediction is performed using the left side of the block, the prediction mode is 1). .

There are three methods of inter prediction: first, through motion estimation, second through merge, and third through skip.

First, the motion estimation method finds a block (prediction block) having the smallest difference from the current block to find the prediction block within the search range, and transmits the residual signal that is the difference between the current block and the prediction block. to be. At this time, the motion parameter indicating the position of the prediction block is transmitted together. The motion parameter includes a motion vector indicating a position of a prediction block, a reference picture index indicating which picture the motion vector is found (indicated by the previous picture or two pictures). ), And a prediction direction flag indicating whether the prediction is performed in the past picture or the future picture (eg, L0: prediction in the past picture, L1: prediction in the future picture). Here, the reference picture index and the prediction direction flag can be encoded with a small number of bits, but in the case of a motion vector, the value itself can have a very large value, which may require a large number of bits for transmission. Therefore, the predictive motion vector is extracted through the prediction of the motion vector separately, and only the residual vector, which is the difference between the current motion vector and the predictive motion vector, is transmitted. The motion vector is predicted to obtain the predictive motion vector (MVP) and the motion of the current block Difference from Vector Only vector is sent. Here, a method of predicting a motion vector is called AMVP (Advanced Motion Vector Prediction).

Secondly, there is a way to merge. Merge uses the motion parameters of the current block and neighboring neighboring blocks as they are for encoding the current block, generates a prediction block using the motion parameters, and converts the residual signal, which is the difference between the current block and the prediction block, to the decoding device. Will be sent. The merge method differs from the AMVP method in that the merge does not directly estimate the motion, but generates the prediction block using the motion parameters of the neighboring blocks. Merger's advantage is that it does not use motion estimation, so the complexity can be reduced, and it is only necessary to send the index of which block the motion parameter is used without transmitting the motion parameter to the decoder, not the motion parameter, but only the index. As the number of data to be transmitted decreases, the coding efficiency is improved.

Third, there is a method called skip. The skip method is similar to the merge method in that the prediction block is generated using the motion parameters of the neighboring block adjacent to the current block and the index of which block the motion parameter is used is different from the merge method. The residual signal for the current block is not encoded.

Hereinafter, the AMVP according to the prior art will be described in more detail. The AMVP is an index indicating a prediction motion vector (MVP), which is a prediction value of a motion vector of a current block, using motion vectors of blocks neighboring the current block, and using a motion vector of a block, which indicates an index indicating a prediction value of a motion vector. It is a method of encoding. The image encoding apparatus calculates a differential motion vector corresponding to the difference between the prediction motion vector and the current motion vector, encodes the same, and transmits the same to the decoding apparatus.

1 is a diagram illustrating neighboring blocks A0, A1, B0, B1, and B2 of a current block used for AMVP.

In Fig. 1, these neighboring blocks used for AMVP are called candidate blocks. When candidate blocks A0 and A1 adjacent to the lower left of the current block are referred to as group A, and candidate blocks B0, B1 and B2 adjacent to the upper side of the current block are referred to as group B, movement of candidate blocks belonging to group A is performed. One of the vectors and one of the motion vectors of the candidate blocks belonging to the B group are selected as the representative motion vector of each group. Temporally, one motion vector is selected as the representative motion vector. An example of a temporal motion vector is a motion vector of a co-located block. A co-located block refers to a block at the same position as the current block in a previous frame.

2 is a diagram illustrating similar position blocks A to E for a temporal representative motion vector.

In FIG. 2, in case of both skip, merge, and AMVP, a temporal candidate is determined by determining a block including one of blocks A to D as a co-located block in a previous frame of the current block. Can be used as As another example of a temporal candidate, block E may be used, and in addition to block E, all neighboring blocks may be selected as temporal candidates.

Among the plurality of representative motion vectors described above, a differential motion vector with a current motion vector and an index indicating which block a motion vector is selected are encoded. The motion vector of the block having the highest coding efficiency is the final predicted motion. Determined by a vector. In the AMVP according to the prior art, as shown in FIG. 1, the number of candidate blocks used for determining the predictive motion vector is limited and predetermined, thereby improving the coding efficiency.

In order to solve the above problem, the present invention encodes by widening the selection of candidate blocks and representative blocks using various neighboring blocks in predicting the motion vectors of the current block more accurately using the motion vectors of neighboring blocks of the current block. Its main purpose is to increase the likelihood of improving efficiency.

According to an embodiment of the present invention, in the apparatus for encoding an image, among the neighboring blocks encoded by using inter prediction, the representative block is selected according to the priority of the motion compensation direction for each neighboring block, and the selected block is selected. And an entropy encoding unit encoding a motion vector of the current block by using a motion vector selected from a motion vector of a representative block and a candidate motion vector of a predetermined neighboring picture as a predictive motion vector. The entropy encoding unit may set the priority of the motion compensation direction with respect to the past direction and the future direction, and select a block in which the motion compensation is performed in the direction having the highest priority among the neighboring blocks as the representative block.

According to an embodiment of the present invention, in the apparatus for decoding an image, the AMVP index is extracted from the bitstream and the priority of the motion compensation direction performed in each neighboring block among the neighboring blocks decoded by using inter prediction. A bitstream decoder configured to select one or more representative blocks according to the at least one representative block and identify the extracted AMVP index in consideration of the selected representative block and candidate motion blocks of a predetermined neighboring picture; And reconstructing the motion vector of the block identified according to the AMVP index among the selected representative block and the candidate motion blocks of the neighboring picture to the prediction motion vector of the current block, and reconstructing the motion vector of the current block to generate a prediction block. It provides a video decoding apparatus comprising a prediction unit. The bitstream decoder may set the priority of the motion compensation direction with respect to the past direction and the future direction, and select a block that performs motion compensation in a direction having a high priority among the neighboring blocks as a representative block. The representative block may be selected one by one from the left neighboring block group and the upper neighboring block group of the current block.

Here, the entropy coding unit may select the representative block by further considering the priority of the distance to the picture referred to by each neighboring block. Further, the priority considered further may be determined depending on whether the distance between the picture referenced by the current block and the picture referenced by the neighboring block is equal to each other.

According to an embodiment of the present invention, in an apparatus for decoding an image, an AMVP index is extracted from a bitstream, and a distance from a picture referenced by each neighboring block for motion compensation among neighboring blocks decoded using inter prediction. A bit for selecting the representative block according to the priority of any one of an inter prediction method, an inter prediction direction, and a block size, and identifying the extracted AMVP index in consideration of the candidate motion block of the selected representative block and a predetermined neighboring picture. A stream decoder; And reconstructing the motion vector of the block identified according to the AMVP index among the selected representative block and the candidate motion blocks of the neighboring picture to the prediction motion vector of the current block, and reconstructing the motion vector of the current block to generate a prediction block. It provides a video decoding apparatus comprising a prediction unit.

According to an embodiment of the present invention, a method of encoding an image comprises: selecting a representative block according to a priority of a motion compensation direction for each neighboring block among neighboring blocks encoded by using inter prediction; And encoding a motion vector of the current block by using a motion vector selected from a motion vector of the selected representative block and a candidate motion vector of a predetermined neighboring picture as a predictive motion vector. do.

According to an embodiment of the present invention, a method of decoding an image, the method comprising: extracting an AMVP index from a bitstream; Selecting one or more representative blocks according to the priority of the motion compensation direction performed in each neighboring block among the neighboring blocks decoded using the inter prediction; Identifying the extracted AMVP index in consideration of the selected representative block and candidate motion blocks of a predetermined neighboring picture; Generating a prediction block by reconstructing a motion vector of a block identified according to the AMVP index among the selected representative blocks and candidate motion blocks of the neighboring picture to a prediction motion vector of the current block, and reconstructing the motion vector of the current block. It provides a video decoding method comprising a. Here, in the step of selecting the one or more representative blocks, the priority of the motion compensation direction for the past direction and the future direction is set, and the block that performs the motion compensation in the direction of high priority among the neighboring blocks as the representative block The representative block may be selected one by one from the left neighboring block group and the upper neighboring block group of the current block.

According to an embodiment of the present invention, in a method of decoding an image, an AMVP index is extracted from a bitstream, and distance from a picture referenced by each neighboring block for motion compensation among neighboring blocks decoded using inter prediction. Selecting a representative block according to the priority of any one of an inter prediction method, an inter prediction direction, and a block size; Identifying the extracted AMVP index in consideration of the selected representative block and candidate motion blocks of a predetermined neighboring picture; Generating a prediction block by reconstructing a motion vector of a block identified according to the AMVP index among the selected representative blocks and candidate motion blocks of the neighboring picture to a prediction motion vector of the current block, and reconstructing the motion vector of the current block. It provides a video decoding method comprising a.

According to the present invention, in predicting the motion vector of the current block more accurately using the motion vectors of the neighboring blocks of the current block, it is possible to improve the coding efficiency by widening the selection of the candidate block and the representative block using the various neighboring blocks. The height is effective.

2 is a diagram illustrating a similar position block for a temporal representative motion vector.

3 is a block diagram schematically illustrating a configuration of an image encoding apparatus according to an embodiment of the present invention.

4 is a diagram illustrating an example of division of a largest coding unit block.

5 is a diagram illustrating an example of a prediction unit block.

FIG. 6 is a diagram illustrating a case of a process of creating a prediction block using pixel values above a current block.

7 is a diagram illustrating the types of intra prediction modes and their indices.

8 is a diagram illustrating an interior of an inter prediction unit for performing inter prediction of various methods.

FIG. 9 is a diagram illustrating a case where a block most similar to a current block is determined as a prediction block in a reference picture through motion estimation.

FIG. 10 is a diagram illustrating an example of neighboring blocks used for inter prediction using a merge method.

FIG. 11 is a diagram illustrating pixels of a reference picture stored in a memory and pixels that are interpolated to less than an integer pixel.

12 is a diagram illustrating a zigzag scan of quantized coefficient blocks.

FIG. 13 is a diagram illustrating a truncated code that encodes an index used for AMVP, merge and skip.

14 is a diagram illustrating an inter prediction mode of a neighboring block of a current block.

15 is a diagram illustrating an inter prediction direction of a neighboring block of a current block.

16 is a diagram illustrating a block size of a neighboring block of the current block.

17 is a flowchart illustrating a method of obtaining a representative motion vector from a motion vector of a candidate block.

18 is a diagram illustrating an inter prediction mode and a prediction direction of neighboring blocks of a current block.

19 shows another example of a neighboring block.

20 is a block diagram schematically illustrating a configuration of an image decoding apparatus according to an embodiment of the present invention.

Hereinafter, a technique for more accurately finding a prediction motion vector most similar to the motion vector of the current block when setting a neighboring candidate block of the AMVP according to the characteristics of the image will be described. EMBODIMENT OF THE INVENTION Hereinafter, embodiment of this invention is described, referring drawings, an equation, etc. Hereinafter, an apparatus for encoding and decoding an image in units of blocks will be described as an example.

3 is a block diagram schematically illustrating a configuration of an image encoding apparatus 100 according to an embodiment of the present invention. Hereinafter, each component of the image encoding apparatus 100 according to an embodiment of the present invention will be described briefly.

The image encoding apparatus 100 according to an embodiment of the present invention is an apparatus for encoding an image, and the image encoding apparatus 100 is largely divided into a block splitting unit 101 and an intra predictor 102. Inter Predictor 103, Transformer 104, Quantizer 105, Motion Compensator 106, Entropy Coder 107, Inverse Quantizer An inverse quantizer 109, an inverse transformer 109, a memory 110, a subtractor 111, an adder 112, and an adder 112 may be included. The video encoding apparatus 100 may be a personal computer (PC), a TV, a notebook computer, a personal digital assistant (PDA), a portable multimedia player (PMP), or a PlayStation Portable (PSP). PlayStation Portable), a wireless terminal, a digital TV, and the like, a communication device such as a communication modem for communicating with various devices or a wired / wireless communication network, a memory for storing various programs and data for encoding an image, Means a variety of devices including a microprocessor for executing and operating a program.

The block dividing unit 101 divides the input image into coding unit blocks. A coding unit block is the most basic unit that is divided for intra prediction / inter prediction, and is a structure that is repeatedly divided into four blocks of the same size (square). For example, the maximum coding unit block may be set to 64x64 size and the minimum coding unit block may be set to 8x8. 4 is a diagram illustrating an example of division of a largest coding unit block. Each coding unit block includes one or more prediction unit blocks as shown in FIG. 5 according to a prediction type. The prediction unit block is the smallest unit that holds the prediction information. More levels may be used, usually using three-level quadtrees, and in general the maximum depths for luma and chroma are the same. In FIG. 5, reference numeral 201 denotes a case where the coding unit block is used as the prediction unit block as it is. (202), (203), (205) and (206) are cases in which two prediction unit blocks of the same size are included, and (204) is a case in which four prediction unit blocks of the same size are included (207) And (208) include two prediction unit blocks having a ratio of 1: 3. In addition to the example of FIG. 5, the coding unit block may be divided into various shapes.

The intra prediction unit 102 generates a predicted block by using the pixel value in the current picture as the current block. Prediction is performed using information in a picture. A prediction block is generated using pixel values neighboring the current block as shown in FIG. 6, and then the difference value with the current block is encoded. In FIG. 6, the prediction block is generated using pixel values existing above the current block. However, in addition to the pixel value above, the neighboring pixel values around the current block as shown in FIG. 7 are set to intra mode using pixel values of various angles. The prediction block may be generated. Each number illustrated in FIG. 7 illustrates an index of the intra prediction mode. After generating the prediction block, an index indicating which prediction mode is used is encoded.

Meanwhile, when intra prediction is performed, a prediction block may be used after smoothing a neighboring pixel using a smoothing filter. For example, when using the [1, 2, 1] filter, the pixels used for prediction are transformed as in Equation 1.

[Equation 1]

Y [n] = (X [n-1] + 2 * X [n] + X [n + 1] + 2) / 4

In Equation 1, X [n] denotes a filtering target pixel, and X [n-1] and X [n + 1] denote left and right adjacent pixels of the filtering target pixel in a direction parallel to the boundary line of the current block.

In addition, whether or not to perform smoothing may be determined according to the size of the prediction unit block and the angle of the prediction mode used for intra prediction. In this case, a look up table may be created and used to determine such smoothing.

8 is a diagram illustrating an inter prediction unit 103 that performs inter prediction of various methods.

The inter prediction unit 103 generates a prediction block by using information of pictures that are previously encoded and decoded in the current block. As illustrated in FIG. 8, prediction may be performed according to three methods such as skip, merge, and motion estimation.

The motion estimation 301 means predicting using information between pictures. Referring to FIG. 9, a block most similar to a current block in a reference picture is determined as a prediction block through motion estimation, and the position of the prediction block is expressed as a motion vector. The entropy encoding unit 107 encodes a corresponding motion parameter. The motion parameter may include a motion vector (MV), a reference picture index, and a prediction direction. In the case of a motion vector, since the efficiency is inferior, the motion vector may be a prediction motion vector. We generate and encode the difference motion vector, which is the difference value from the original motion vector. The image decoding apparatus to be described later reconstructs the prediction block by using the motion parameter extracted from the bitstream transmitted from the image encoding apparatus, and then decodes the current block by adding the values of the reconstructed prediction block and the residual block extracted from the bitstream.

The AMVP 302 (Advanced Motion Vector Prediction) is a method of predicting the motion vector of the current block using the motion vector of the neighboring block and encoding the index of which block the motion vector is predicted using. Equation 2 shows an equation for calculating a differential motion vector (MVD).

[Equation 2]

MVD = MV-MVP

Here, MV is a motion vector of the current block, MVP is a predictive motion vector predicted by a neighboring block, and MVD is a difference motion vector which is a difference value between the motion vector and the predictive motion vector of the current block.

Merge 303 is one of techniques for determining a motion parameter for a current block in a neighboring block of the current block. Referring to FIG. 10, an optimal block for merging may be found in blocks A through E and the temporal candidate blocks of FIG. 2, or two having higher priority by giving priority to blocks A through E of FIG. 10. An optimal block may be determined among the two candidate blocks (ie, the representative block) and the temporal candidate blocks in FIG. 2. For reference, in all embodiments, a temporal candidate block means a block of pictures that are located in the temporal vicinity, and a block located in the immediately preceding picture but present in the same spatial position as the current block may be used. The meaning of the temporal candidate block is the same as that of the merge even when used for AMVP.

Skip 304 refers to a case where a residual signal value of the residual block is all zero when a block (prediction block) most similar to the current block is found using the motion parameter of the neighboring block. In this case, the prediction block is made using the motion parameters of the neighboring blocks without encoding the residual signal separately, and the current block is encoded. In order to indicate which block motion parameter information is used, the index of the corresponding block is encoded.

Candidates of neighboring blocks neighboring the current block for AMVP, merge, and skip are candidate blocks other than neighboring blocks A0 to B2 shown in FIG. 1 and neighboring blocks A to E shown in FIG. Can be.

The subtraction unit 111 generates a residual block by subtracting the prediction block from the current block.

The transform unit 104 generates a transform block by transforming the residual block. The transform block is a unit used for the transform and quantization process. The transform unit may be divided in the same manner as the coding unit as shown in FIG. 4, or may be transformed by splitting in other various ways. Similar to the coding unit block, the information on the transformation unit may use a quadtree structure, and the transformation unit may have various sizes. The transform unit 104 converts the residual signal into the frequency domain to generate and output a transform block having a transform coefficient. Here, as a method of transforming the residual signal into the frequency domain, various transform techniques such as Discrete Cosine Transform (DCT) based, Discrete Sine Transform (DST), and Karhunen Loeve Transform (KLT) can be used. Using this, the residual signal is converted into a frequency domain and converted into a transform coefficient. In order to conveniently use the transform method, matrix operations are performed using a basis vector. Depending on how the prediction block is encoded, the transform methods may be mixed and used in the matrix operation. For example, in intra prediction, a discrete cosine transform may be used in the horizontal direction and a discrete sine transform in the vertical direction.

The quantization unit 105 quantizes the transform block to generate a quantized transform block. That is, the quantization unit 105 quantizes the transform coefficients of the transform block output from the transform unit 104 to generate and output a quantized transform block having a quantized transform coefficient. Here, dead zone uniform threshold quantization (DZUTQ) or quantization weighted matrix (DZUTQ) may be used as the quantization method, but various quantization methods such as quantization improved therefrom may be used.

The inverse quantization unit 108 performs inverse quantization by inversely performing a quantization technique used for quantization on a quantized transform block and restores an inverse quantized transform block having a transform coefficient.

The inverse transform unit 109 inversely transforms the inverse quantized transform block to restore the residual block having the residual signal by using the same method as the transform method. The inverse transform is performed by inversely performing the transform technique used by the transform unit 104. .

The adder 112 adds a residual signal reconstructed by the inverse transform unit 109 and a prediction image generated through intra prediction or inter prediction to reconstruct the current block.

The memory 110 stores the current block reconstructed by adding the residual signal reconstructed by the inverse transform unit 109 and the prediction image generated through intra prediction or inter prediction, and predicts another block such as the next block or the next picture. Can be utilized.

The motion compensator 106 may perform motion compensation after increasing the resolution of the reference picture through interpolation between pixel values of the reference picture for pixels smaller than an integer pixel.

11 illustrates pixels of a reference picture stored in a memory and pixels interpolated to less than an integer pixel. Filtering the previously reconstructed pixels A to U of the reference picture with an interpolation filter can generate interpolated pixels a to s, and interpolated pixels a to s between the already reconstructed pixels as shown. The resolution of the reference picture may be four times higher than interpolation. In this case, various filters such as a bilinear filter, a DCT based filter, and a cubic interpolation filter may be used.

The entropy encoding unit 107 encodes the quantized transform block and outputs a bitstream. That is, the entropy encoding unit 107 entropy encodes a frequency coefficient string obtained by scanning the quantized transform coefficients of the quantized transform block output from the quantization unit 105 by various scan methods such as a zigzag scan. Bitstream including additional information (for example, information about a prediction mode, quantization coefficient, motion parameter, etc.) necessary for encoding using various encoding techniques, and decoding the corresponding block in an image decoding apparatus to be described later. Create and print

FIG. 13 is an example of a truncated code that encodes an index used for AMVP, merge, and skip. Bits may be allocated differently according to the number of blocks available nearby. In addition, it is possible to use only a table for the maximum number of blocks, as shown in the first table of FIG. 13, regardless of the number of available blocks nearby.

In the present embodiment, the entropy encoding unit 107 selects a representative block according to the priority of the inter prediction method performed from the neighboring blocks that have performed the inter prediction, and selects a representative block from the motion vectors of the selected representative block and the blocks included in the previous picture. The motion vector of the current block can be encoded by using the optimal motion vector most similar to the motion vector of the current block among the motion vectors of the block located at the position similar to the current block as the prediction motion vector. Here, the inter prediction method means one of an AMVP method, a skip, and a merge. In addition, a block located at the same or similar position as the current block among the blocks included in the previous picture means a temporal candidate block as described above. The entropy encoding unit 107 selects an optimal motion vector most similar to the current motion vector among the selected representative motion vectors as a predictive motion vector, encodes an index of the selected position, and transmits the index to the image decoding apparatus. In addition, the entropy encoding unit 107 encodes the difference value between the current motion vector and the set prediction motion vector, generates a bitstream, and transmits the bitstream to the image decoding apparatus.

In the AMVP according to the related art, only motion vectors of a predetermined number of blocks, as shown in FIG. 1, among neighboring blocks of the current block are used for representative motion vector selection. However, in the present embodiment, unlike the prior art, more motion vectors of more neighboring blocks (eg, motion vectors of all neighboring blocks) are used for representative motion vector selection.

In the first to third embodiments to be described below, when the AMVP 302 is used, the representative block for the AMVP in the neighboring candidate blocks neighboring to the current block according to the inter prediction mode is performed in the left group and the upper two groups. It shows how to select one each and set it as a representative block for AMVP. For reference, in the following description, the case where the representative vector is selected from the candidate blocks means that the representative block is selected from the candidate blocks according to a predetermined criterion and the motion vector most similar to the current motion vector among the motion vectors of the selected representative block. (Ie, representative vector) is set as a predicted motion vector.

(First embodiment)

In the present embodiment, the entropy encoding unit 107 may include two groups (eg, left block groups (a, b, c) and an upper block in FIG. 14 to which neighboring candidate blocks belong when encoding information on a motion vector of the current block). Group (d, e, f, g)), respectively, to select a representative block, a total of three including the motion vector of the two selected representative blocks and the motion vector (that is, the temporal candidate vector) of the temporal candidate block The most optimal predictive motion vector is selected among the representative vectors and the corresponding index is assigned and encoded. The residual vector mvd, which is a difference value between the motion vector of the original current block and the selected predicted motion vector, is encoded.

If the priority of the inter prediction mode is set to 'AMVP> Skip> Merge', in the process of selecting a representative vector, a block encoded using AMVP is preferentially selected, and if there is no block using AMVP, skipping, The blocks encoded in the merge order are selected. Therefore, in FIG. 14, block c is selected as the representative vector in the left group and block f in the upper group. In FIG. 14, if there are a plurality of blocks having the same inter prediction mode, the representative vector is given priority in the selection direction such as 'right> left' in the upper group and 'down> up' in the left group. Select. Here, the priority of the selection direction may be the opposite direction. In addition, generation of the representative vector is not necessarily limited to one in the left group and one in the upper group, and a plurality of representative vectors may be selected in one group. In addition, the priority of the inter prediction mode is not limited to 'AMVP> skip> merge', and the priority may be determined in various ways for the AMVP, skip, and merge methods.

On the other hand, in the case of decoding by the video decoding apparatus to be described later, the process of selecting the representative block is the same as the process of selecting the representative block in the video encoding process. In addition, in an image decoding apparatus to be described later, an index of which block is finally selected from a representative block and a temporal candidate block is extracted from a bitstream, and a motion vector of a block corresponding to the extracted index is decoded into a prediction motion vector, The residual vector decoded from the stream is added to reconstruct the motion vector of the final current block.

Second Embodiment

In the second embodiment, the entropy encoding unit 107, like the first embodiment, includes two groups to which neighboring candidate blocks belong when encoding information on the motion vector of the current block (left groups (a, b, c) and the uppermost group (d, e, f, g)) select the most optimal predictive motion vector among the representative vectors of the three blocks, including the representative block and the temporal candidate block, respectively, and assign the corresponding index. To encode it. The residual vector mvd, which is a difference value between the motion vector of the original current block and the selected predicted motion vector, is encoded. The same applies to the third embodiment described later.

However, the second embodiment is different from the first embodiment in that the representative block is selected according to the priority of the inter prediction direction of the neighboring blocks. In the example of FIG. 15, after confirming that neighboring blocks of the current block are encoded in one direction or two directions, and if priority is set (for example, 'unidirectional> bidirectional'), the motion vector of the block encoded in one direction is preferentially represented as a representative vector. Can be selected. As in the first embodiment, the representative block may select one representative block from the left group block (a, b, c, d) and the upper group block (d, e, f, g). One or more representative blocks may be selected in each of the and upper groups, or only one representative block may be selected in the left group and the upper group. The priority of the prediction direction may be set to two or more of the three directions of the past direction (list0), the future direction (list1), and the bidirectional direction.

If the priority of 'bidirectional> unidirectional' is set, the priority of the selection direction of 'down> up' is set in the left group, and the priority of selection direction of 'right> left' is set in the upper group. In this case, block a is selected from the left group and block g is selected from the upper group as the representative block in FIG. 15.

In addition, one or more representative blocks may be selected from the candidate blocks in consideration of the priority of the motion compensation direction of the candidate block according to the motion compensation direction of the current block. For example, if the motion estimation of the current block is progressing in the past direction, the motion compensation direction is the same as the motion estimation direction of the current block with respect to one of the surrounding blocks (ie, the first candidate block). Determine whether to select the representative block. If it is different from the motion estimation direction of the current block with respect to the first candidate block, it is determined whether the serial number (POC: Pictute of Count) of the reference picture is identical with each other for the current block and the first candidate block. If the numbers are the same, the first candidate block may be selected as the representative block.

If the first candidate block is not selected as the representative block after checking whether to select the representative block in consideration of the priority of the motion compensation direction with respect to the first candidate block as described above, another candidate block (ie, the second candidate block) Similarly to the case of the first candidate block, it is determined whether to select the second candidate block as a representative block in consideration of the priority of the motion compensation direction.

(Third Embodiment)

In the third embodiment, the entropy encoding unit 107 selects the representative block according to the priority of the block size of the neighboring block, which is different from the first and second embodiments. In the example of FIG. 16, when the priority of the block size is set (for example, 'large block> small block'), it is checked whether the neighboring blocks of the current block are large or small, and the block encoded into the large block is represented as a representative block. Can be selected first.

If the priority of 'big block> small block' is set, the priority direction of 'down> up' is set in the left group, and the priority direction of 'right> left' in the upper group. In this case, blocks b and d are respectively selected as the representative blocks in FIG. 16.

In addition, it is also possible to select a representative block according to the priority of the size of the neighboring block to the size of the current block. For example, when the size of the current block is set to 16 × 16, a high priority may be set to a block near the size of 16 × 16 among the neighboring blocks and selected as the representative block among the neighboring blocks. The motion vector of the representative block selected here becomes a representative motion vector (ie, a representative vector).

(Example 4)

In the fourth embodiment, the entropy encoding unit 107 selects one or more representative blocks from the neighboring blocks according to the priority of the distance of the pictures referred to from the neighboring blocks, which is different from the first, second, and third embodiments. to be.

In the example of FIG. 16, when the distance between the reference picture is different for each neighboring block, the distance between the neighboring block and the referenced picture is set in order of priority (for example, the distance from the reference picture of the neighboring block c is 2, neighboring). When the distance from the reference picture of the block d is 1 and the distance from the reference picture of the neighboring block e is 3, the highest priority is set in the neighboring block d) so that the motion vector of the nearest block can be set as the representative vector.

It is also possible to compare with the distance of the picture referred to by the current block. For example, if the distance from the picture referenced in the motion estimation process of the current block is 2, the high priority is set to the neighboring block whose distance from the picture referenced by the corresponding block among the neighboring blocks is closest to 2 to be the representative block. The motion vector may be selected as a representative vector. In addition, in selecting the representative block, the priority block may be determined according to the motion compensation direction, and then the representative block may be selected in consideration of the priority of the distance to the picture referenced by each neighboring block. In this case, the priority may be determined according to whether the distance between the picture referenced by the current block and the picture referenced by the neighboring block is equal to each other when the priority is the same according to the motion compensation direction. That is, if the distances are equal to each other, the priority is high, and if the distances are not equal, the priority may be set to be low.

Meanwhile, in the AMVP according to the related art, a predetermined number of blocks among neighboring blocks of the current block are used as candidate blocks for selecting a representative block. However, in the fifth to eighth embodiments, unlike the prior art, there is a difference in that candidate blocks are selected from among neighboring blocks according to a predetermined criterion.

Next, the fifth to eighth embodiments of the present invention will be described.

In the fifth to eighth embodiments, the entropy encoding unit 107 selects a candidate block according to a predetermined priority from among neighboring blocks on which the inter prediction is performed, and represents the representative block according to a predetermined condition among the selected candidate blocks. Next, the motion vector of the current block is encoded by using the optimal motion vector as the predictive motion vector among the motion vectors of the selected representative block and the motion vectors of the temporal candidate blocks.

It is assumed here that the number of candidate blocks selects two candidate blocks from the left group of neighboring blocks and three candidate blocks from the upper group. In addition, the priority of the selection direction in each group is assumed to be 'down> up' in the left group and 'right> left' in the upper group unless otherwise stated.

(Example 5)

In the fifth embodiment, the entropy encoding unit 107 selects a candidate block according to the priority of the performed inter prediction method among neighboring blocks on which the inter prediction is performed.

There is a difference from the prior art in the method of selecting the candidate block and the representative block (representative vector). In the first to fourth embodiments, the representative vector is immediately selected according to a predetermined criterion in each group. However, in the present embodiment, a predetermined number of candidate blocks is determined for each group according to the inter prediction mode, and then representative blocks are selected using FIG. 17.

In the example of FIG. 18, when priority is set to 'AMVP> skip> merge', for example, in the left group (a, b, c), block c using AMVP and block a using skip are representative blocks of FIG. 17. The final candidate block to be used in the configuration algorithm is selected. Similar to the left group, in the upper group d, e, f, g, blocks g using AMVP, block e using d and skip are selected as the final candidate blocks to be used in the representative block setting algorithm of FIG. The number of candidate blocks selected from each group may vary depending on the embodiment, and the priority of the inter prediction method may also vary. Also, the priority may be determined by mixing the inter prediction mode and the motion compensation direction.

After the candidate blocks are selected in each group, the representative vector is selected using the method of FIG. Hereinafter, a process of determining the representative vector of the group A will be described with reference to FIG. 17. Here, the prediction direction may be used as one kind of motion compensation direction (past or future). Assume that candidate blocks of group A are A0 (block 1) and A1 (block 2). At this time, A0 becomes block 1 in FIG. 17 and A1 becomes block 2. FIG. First, start with X = 1 (S1701). It is determined whether the reference picture index and the prediction direction of the block X are the same as the current block (S1702). If the determination of step S1702 is YES, block X becomes a representative block (ie, the motion vector of the block becomes a representative vector). If the determination result of step S1702 is no, it is determined whether there is a remaining block (S1703). If the determination result of step S1703 is YES, 1 is increased to X (ie, block 2). The flow proceeds to step S1702 (S1704).

If the determination result in step S1703 is no, 1 is substituted into X (S1705), and it is checked whether the reference picture index of block X is the same as the current block but the prediction direction is different (S1706). If the determination of step S1706 is YES, block X becomes a representative block. If the determination result of step S1706 is no, the reference picture index of block X is different from the current block, but it is determined whether the prediction direction is the same (S1707). If the determination result of step S1707 is YES, block X becomes a representative block. If the determination result of step S1707 is no, it is checked whether the reference picture index of the block X and the prediction direction are different from the current block (S1708). If the result of the determination of step S1708 is YES, block X becomes a representative block. If the result of the determination in step S1708 is no, it is determined whether there are more remaining blocks. If there is a remaining block, X is incremented to 1 (that is, block 2) and the process proceeds to step S1706 again (S1710).

In this case, when both are determined to be “no” in step 1708, the present invention means that the block is not available for inter prediction. For example, this is the case of an intra prediction block.

Therefore, the representative vector can be extracted in a similar way for the upper group.

(Example 6)

In the sixth embodiment, the entropy encoding unit 107 selects a candidate block according to the priority of the performed inter prediction direction among neighboring blocks on which the inter prediction is performed.

If the priority is set to 'unidirectional> bidirectional', the unidirectional blocks among neighboring blocks are preferentially set as candidate blocks, and then a representative vector is selected using the algorithm of FIG. In addition, it is also possible to select a candidate block by setting priorities for two or more of the past direction (list0), future direction (list1), both directions.

Therefore, in the case of FIG. 18, when the priority is set to 'unidirectional> bidirectional', block c and block a are selected as candidate blocks in the left group, and block f, block d and block g are candidate blocks in the upper group. Is selected. Here, one representative block for each group is selected for each candidate block for each group according to the algorithm of FIG. 17.

It is also possible to compare with the motion compensation direction of the current block. For example, if the motion compensation of the current block is progressing in the past, the motion compensation direction of the neighboring blocks is set to a candidate block by setting a high priority to a block having the past motion compensation direction and then using the algorithm of FIG. 17 as a representative vector. You can also choose.

(Example 7)

In the seventh embodiment, the entropy encoding unit 107 selects a candidate block according to the priority of the performed inter prediction block size among neighboring blocks on which the inter prediction is performed.

If the priority is set to 'large block> small block', the blocks of the large block are preferentially set as candidate blocks, and then the representative vector is selected using the algorithm of FIG.

Therefore, in the case of FIG. 16, when the priority is set to 'big block> small block', block b and block a are selected as candidate blocks in the left group, and block d, block e and block c are selected in the upper group. It is selected as a candidate block. Here, one representative block for each group is selected for each candidate block for each group according to the algorithm of FIG. 17.

It is also possible to compare with the size of the current block. For example, when the size of the current block is set to 16 × 16, a high priority is set to a block close to 16 × 16 among the neighboring blocks and set as a candidate block, using the algorithm of FIG. You can also choose.

(Example 8)

In the eighth embodiment, the entropy encoding unit 107 selects a candidate block according to the priority of the distance to the referenced picture among the neighboring blocks on which the inter prediction is performed.

If the priority is set to 'short distance> long distance', the priority of the reference picture is as follows: the distance from the reference picture of the neighboring block c is 2, the distance of the reference picture of d is 1, When the distance of the reference picture of e is 3, the highest priority is set to d), and the block closest to the distance of the reference picture is preferentially set as the candidate block, and then the representative vector is selected using the algorithm of FIG.

It is also possible to compare with the distance of the picture referenced by the current block. For example, when the distance from the reference picture is 2 during the motion estimation process of the current block, the priority block is set after the high priority is set to a block closest to 2 among the neighboring blocks. The algorithm of 17 may be used to select a representative vector.

Meanwhile, in applying the representative block selection algorithm of FIG. 17, the candidate blocks are candidates in the search order according to any priority among the inter prediction method, the inter prediction direction, the motion compensation direction, the block size, and the distance to the reference picture. It is determined whether a block meets a predetermined condition and is selected as a representative block. For example, if the priority is set to 'unidirectional> bidirectional', two 17 algorithms are applied with a unidirectional block to block 1 and a bidirectional block to block 2. If there are two unidirectional blocks, the lower block becomes block 1 and the upper block becomes block 2 according to the selection direction priority (for example, 'down> up'), and the algorithm of FIG. 17 is applied.

On the other hand, Figure 19 shows another example of the peripheral block. If a candidate block is selected in the fifth embodiment, candidate blocks for each group are found in the order of AMVP, skip, and merge among the interpredicted blocks. Thus, two candidate blocks are selected in the left group and three in the upper group.

The image decoding apparatus 400 according to an embodiment of the present invention is a personal computer (PC), a notebook computer, a TV, a personal digital assistant (PDA), a portable multimedia player, as the above-described image encoding apparatus. (PMP: Portable Multimedia Player), PlayStation Portable (PSP: PlayStation Portable), wireless terminal (Wireless Terminal), digital TV, etc., communication devices such as a communication modem for communicating with various devices or wired or wireless communication network, video Means a variety of devices including a memory for storing the program and a memory for storing the data, a microprocessor for the operation and control by running the program.

The image decoding apparatus 400 according to an embodiment of the present invention includes a bitstream decoder 401, an inverse quantizer 402, an inverse transformer 403, a predictor 405, an adder 409, and an adder. It may be configured to include a memory 408.

The bitstream decoder 401 extracts a quantized transform block by decoding the bitstream. That is, the bitstream decoder 401 decodes and inversely scans the bit stream extracted from the input bitstream to restore the quantized transform block having the quantized transform coefficients. In this case, the bitstream decoder 401 may decode using an encoding technique such as entropy encoding used by the entropy encoder 107.

In the inter prediction, the bitstream decoder 401 extracts and decodes the encoded difference vector from the bitstream, reconstructs the differential vector, and extracts an AMVP index for the differential motion vector used in the AMVP from the bitstream. Here, the motion vector of the neighboring block corresponding to the position identified from the extracted AMVP index is reconstructed into the predicted motion vector, and the reconstructed predictive motion vector and the decoded difference vector are added to reconstruct the motion vector of the current block to be reconstructed.

Meanwhile, the process of selecting a representative block in order to decode the motion vector in the bitstream decoder 401 of the image decoding apparatus 400 according to an embodiment of the present invention may include: a bitstream decoder of the image encoding apparatus 100. It is the same as the representative block selection process in the encoding process performed at 401. Accordingly, the bitstream decoding unit 401 selects the representative block according to at least one priority from among the inter prediction method, the motion compensation direction, the block size, and the distance to the referenced picture with respect to predetermined neighboring blocks of the current block. .

In addition, in the case of intra prediction, the bitstream decoder 401 extracts and decodes an encoded intra prediction mode index from the bitstream to inform which intra prediction mode the current block used.

The inverse quantization unit 402 dequantizes the quantized transform block. That is, the inverse quantization unit 402 inverse quantizes the quantized transform coefficients of the quantized transform block output from the bitstream decoder 401. In this case, the inverse quantization unit 402 reversely quantizes the quantization technique used by the quantization unit 105 of the image encoding apparatus.

The inverse transform unit 403 inversely transforms the inverse quantized transform block output from the inverse quantization unit 402 to restore the residual block. That is, the inverse transformer 403 restores the residual block having the residual signal reconstructed by inversely transforming the inverse quantized transform coefficients of the inverse quantized transform block output from the inverse quantizer 402. Inverse transformation is performed by performing the transformation technique used in ().

The prediction unit 405 may include an intra prediction unit 406 and an inter prediction unit 407, and the intra prediction unit 102 and the inter prediction unit 103 of the image encoding apparatus described above with reference to FIG. 3, respectively. It has a similar function.

The adder 409 adds the inverse transformed residual signal and the prediction image generated through intra prediction or inter prediction to reconstruct the current block image.

The memory 408 may store the decoded image in the same manner as the memory of the image encoding apparatus and may be used for later prediction.

In the image decoding apparatus, the process of selecting the representative block is the same as in the image encoding apparatus, and the bitstream decoding unit 401 extracts an index from the bitstream and decodes which block was finally selected as the differential motion vector. The motion vector and the decoded residual vector are added to restore the final motion vector of the current block.

The intra prediction unit 406 uses the motion vector generated here to predict the current block to be reconstructed.

Meanwhile, the image encoding / decoding apparatus according to an embodiment of the present invention may be implemented by connecting a bitstream (encoded data) output terminal of the image encoding apparatus of FIG. 3 to a bitstream input terminal of the image decoding apparatus of FIG. 20.

An image encoding / decoding apparatus according to an embodiment of the present invention includes any one of an inter prediction method, an inter prediction direction, a motion compensation direction, a block size, and a distance from a reference picture with respect to predetermined neighboring blocks of the current block. Select the representative block according to the rank or select the candidate block according to the priority of any one of the inter prediction method, the inter prediction direction, the motion compensation direction, the block size, and the distance to the reference picture with respect to a predetermined neighboring block of the current block. The representative block is selected according to a predetermined condition with respect to the selected candidate block, and the optimal motion vector is selected from the motion vector of the selected representative block and the motion vector of the block of another picture at a position similar to the current block (ie, a temporal candidate motion vector). Video encoding apparatus 100 for encoding the motion vector of the current block into a bitstream using Implementing an image encoder in an image encoding / decoding apparatus according to an embodiment of the present invention, extracting an index (AMVP index) and a residual vector for a prediction motion vector from a bitstream, and encoding the neighboring block using inter prediction Among them, the representative block corresponding to the index is identified by selecting the representative block according to at least one priority among the inter prediction method, the inter prediction direction, the motion compensation direction, the block size, and the distance from the reference picture. An image decoding apparatus 400 for reconstructing a motion vector of a block into a predicted motion vector of a current block and reconstructing the current block using the reconstructed predictive motion vector (image in an image encoding / decoding apparatus according to an embodiment of the present invention). May be implemented).

In an image encoding method according to an embodiment of the present invention, the image decoding apparatus 100 divides an input image into coding unit blocks by the block division unit 101, and the intra prediction unit 102 selects a current block. Generating a prediction block using pixel values in a picture or inter prediction unit 103 generating a prediction block using information of previously encoded and decoded pictures of the current block, and subtracting unit 111 in the current block. Generating a residual block by subtracting the prediction block, transforming the residual block to generate a transform block, generating a transform block by quantization 105, and generating a quantized transform block by quantizing the transform block; An entropy encoding step in which the entropy encoding unit 107 encodes a transform block quantized by the quantization unit 105 and outputs it as a bitstream, and inverses the block quantized by the quantization unit 105. Inverse quantization of the magnetization unit 108 to restore a transform block having a transform coefficient, and inverse transform unit 109 inverse transforms a transform block inversely quantized by the inverse quantization unit 108 to restore a residual block having a residual signal Reconstructing the current block by adding the inverse transformed residual signal and the prediction image generated through intra prediction or inter prediction, and the memory 110 inversely transforming the residual signal, intra prediction or the like. And storing the reconstructed current block by adding the prediction image generated through the inter prediction.

In an image encoding method according to an embodiment of the present invention, the first embodiment of the entropy encoding step selects a representative block according to the priority of the used inter prediction method from among neighboring blocks encoded using inter prediction. The motion vector of the current block is encoded by using the motion vector selected from the motion vector and the temporal candidate motion vector of the selected representative block as the prediction motion vector. Here, the method of selecting a representative block according to the priority of the inter prediction method includes setting priorities for the prediction method (AMVP), the skip (SKIP) method, and the merge method using motion estimation, and among the neighboring blocks. It may mean that a predetermined number of blocks on which inter prediction is performed in a high priority method is selected as a representative block.

In the video encoding method according to an embodiment of the present invention, the second embodiment of the entropy encoding step selects a representative block according to the priority of the used inter prediction direction from among neighboring blocks encoded using inter prediction. The motion vector of the current block is encoded by using the motion vector selected from the motion vector and the temporal candidate motion vector of the selected representative block as a prediction motion vector. Here, in the method of selecting a representative block according to the priority of the inter prediction direction, a predetermined number of blocks in which priority for unidirectional and bidirectional prediction are set, and inter prediction is performed in a prediction direction having a higher priority among neighboring blocks. This may mean selecting as a representative block. In addition, the method of selecting a representative block according to the priority of the inter prediction direction, the priority of the two or more prediction directions of the past prediction, future prediction and bidirectional prediction, and sets the priority of the prediction direction with a higher priority among the neighboring blocks This may mean selecting a predetermined number of blocks on which inter prediction is performed as a representative block.

In a video encoding method according to an embodiment of the present invention, the third embodiment of the entropy encoding step selects a representative block according to the priority of the block size from among neighboring blocks encoded by using inter prediction, and selects the selected representative. A motion vector of the current block is encoded using a motion vector selected from a motion vector of the block and a temporal candidate motion vector as a prediction motion vector. Here, in the method of selecting a representative block according to the priority of the block size, the priority of the small block and the large block is set, and a predetermined number of blocks in which inter prediction is performed with a block size having a high priority among neighboring blocks. It may mean selecting as a representative block.

In a video encoding method according to an embodiment of the present invention, the fourth embodiment of the entropy encoding step selects a candidate block according to the priority of the used inter prediction method from among neighboring blocks encoded using inter prediction. The representative block is selected from among the selected candidate blocks according to a predetermined condition, and the motion vector of the current block is encoded using a motion vector selected from the motion vector and the temporal candidate motion vector of the selected representative block as a prediction motion vector. Here, the method for selecting a candidate block according to the priority of the inter prediction method includes setting a priority for a prediction method (AMVP), a skipping method (SKIP) method, and a merging method using motion estimation, and among the neighboring blocks. This may mean selecting a predetermined number of blocks for which inter prediction is performed in a high priority method as candidate blocks.

In a video encoding method according to an embodiment of the present invention, the fifth embodiment of the entropy encoding step selects a candidate block according to the priority of the used inter prediction direction from among neighboring blocks encoded using inter prediction. The representative block is selected from among the selected candidate blocks according to a predetermined condition, and the motion vector of the current block is encoded using the motion vector selected from the motion vector and the temporal candidate motion vector of the selected representative block as a prediction motion vector. Here, in the method of selecting a representative block according to the priority of the inter prediction direction, a predetermined number of blocks in which the priority for unidirectional and bidirectional prediction are set and the inter prediction is performed in a method having a higher priority among neighboring blocks are candidates. Select as a block or select a preset number of blocks that set priorities for two or more prediction directions among past prediction, future prediction, and bidirectional prediction, and perform inter prediction by using a higher priority among neighboring blocks as candidate blocks. It can mean doing.

In a video encoding method according to an embodiment of the present invention, the sixth embodiment of the entropy encoding step may include selecting a candidate block according to a priority of a block size among neighboring blocks encoded by using inter prediction, and selecting the selected block. A representative block is selected according to a predetermined condition among candidate blocks, and a motion vector of the current block is encoded using a motion vector selected from the motion vector and the temporal candidate motion vector of the selected representative block as a prediction motion vector. Here, in the method of selecting candidate blocks according to the priority of the block size, the priority of the small blocks and the large blocks is set, and the predetermined number of blocks in which inter prediction is performed with the block size having the highest priority among the neighboring blocks is performed. This may mean selecting as a candidate block.

An image encoding / decoding method according to an embodiment of the present invention may be realized by combining the image encoding method according to an embodiment of the present invention and the image decoding method according to an embodiment of the present invention.

In the video encoding / decoding method according to an embodiment of the present invention, any one of a priority among an inter prediction method, an inter prediction direction, a motion compensation direction, a block size, and a distance from a reference picture among neighboring blocks on which inter prediction is performed The candidate block is selected according to the priority of any one of the representative blocks or the neighboring blocks that have performed the inter prediction, among the inter prediction method, the inter prediction direction, the motion compensation direction, the block size, and the distance from the reference picture. A representative block is selected for the selected candidate block according to a predetermined condition, and the motion vector of the current block is a bitstream using an optimal motion vector among the motion vector and temporal candidate motion vector of the selected representative block as a prediction motion vector. Video encoding step of encoding; And extracting an index (AMVP index) and a residual vector for a prediction motion vector from the bitstream, and using the inter prediction method, an inter prediction method, an inter prediction direction, a motion compensation direction, a block size, and a reference picture among neighboring blocks encoded using inter prediction. Select the representative block according to the at least one priority among the distances from the target and identify the index in consideration of the selected representative block to restore the motion vector of the representative block corresponding to the index to the predicted motion vector of the current block The video decoding step of recovering the current block by using a.

The above description is merely illustrative of the technical idea of the present invention, and those skilled in the art to which the present invention pertains may make various modifications and changes without departing from the essential characteristics of the present invention. Therefore, the embodiments disclosed in the present invention are not intended to limit the technical idea of the present invention but to describe the present invention, and the scope of the technical idea of the present invention is not limited by these embodiments. The protection scope of the present invention should be interpreted by the following claims, and all technical ideas within the equivalent scope should be interpreted as being included in the scope of the present invention.

According to the present invention, encoding efficiency is increased by widening the selection of candidate blocks and representative blocks using various neighboring blocks in predicting the motion vectors of the current block more accurately by using the motion vectors of neighboring blocks of the current block. This is a very useful invention because it has the effect of increasing the possibility of improving.

CROSS-REFERENCE TO RELATED APPLICATION

This patent application is filed with the Korean Patent Application No. 10-2011-0123841, filed in Korea on November 24, 2011, and Patent Application No. 10-2012-0133206, filed in Korea on November 22, 2012. If priority is claimed under section (a) (35 USC § 119 (a)), all of that is incorporated by reference into this patent application. In addition, if this patent application claims priority for the same reason for countries other than the United States, all its contents are incorporated into this patent application by reference.

Claims

In the apparatus for encoding a video,

Among the neighboring blocks encoded by using inter prediction, a representative block is selected according to the priority of the motion compensation direction for each neighboring block, and the selected motion is selected from the motion vector of the selected representative block and the candidate motion vector of the preset neighboring picture. An entropy coding unit encoding a motion vector of the current block by using a vector as a predictive motion vector.

An image encoding apparatus comprising a.
The method of claim 1,

The entropy coding unit,

And a priority block of a motion compensation direction for the past direction and the future direction, and selecting a block that performs motion compensation in a direction having a higher priority among the neighboring blocks as a representative block.
The method of claim 1, wherein the entropy coding unit,

And the representative block is selected in consideration of the priority of the distance to the picture referenced by each neighboring block.
In the apparatus for decoding an image,

Extract the AMVP index from the bitstream, select one or more representative blocks according to the priority of the motion compensation direction performed in each neighboring block among the neighboring blocks decoded using inter prediction, and select the selected representative block and the predetermined neighbor A bitstream decoder that identifies the extracted AMVP index in consideration of candidate motion blocks of a picture; And

An inter prediction for generating a prediction block by reconstructing a motion vector of a block identified according to the AMVP index among the selected representative blocks and candidate motion blocks of the neighboring picture to a prediction motion vector of a current block, and reconstructing the motion vector of the current block. part

Video decoding apparatus comprising a.
The method of claim 4, wherein

The bitstream decoder,

And setting a priority of the motion compensation direction for the past direction and the future direction, and selecting a block in which the motion compensation is performed in the direction having the highest priority among the neighboring blocks as the representative block.
The method of claim 4, wherein

The representative block,

The video decoding apparatus, characterized in that the selection from the left neighboring block group and the upper neighboring block group of the current block one by one.
The method of claim 4, wherein the entropy coding unit,

And the representative block is selected in consideration of the priority of the distance to the picture referenced by each neighboring block.
The method of claim 7, wherein

And the priority is determined based on whether the distance between the picture referenced by the current block and the picture referenced by the neighboring block is equal to each other.
In the apparatus for decoding an image,

The AMVP index is extracted from the bitstream, and the representative block is selected according to the priority of any one of an inter prediction method, an inter prediction direction, and a block size for each neighbor block among the neighbor blocks decoded using inter prediction. A bitstream decoder that identifies the extracted AMVP index in consideration of the selected representative block and candidate motion blocks of a predetermined neighboring picture; And

An inter prediction for generating a prediction block by reconstructing a motion vector of a block identified according to the AMVP index among the selected representative blocks and candidate motion blocks of the neighboring picture to a prediction motion vector of a current block, and reconstructing the motion vector of the current block. part

Video decoding apparatus comprising a.
In the method of encoding an image,

Selecting a representative block according to the priority of the motion compensation direction for each neighboring block among neighboring blocks encoded by using inter prediction; And

Encoding a motion vector of the current block by using a motion vector selected from a motion vector of the selected representative block and a candidate motion vector of a preset neighboring picture as a prediction motion vector;

Image encoding method comprising a.
The method of claim 10, wherein in the step of selecting the representative block,

And selecting the representative block by further considering the priority of the distance to the picture referenced by each neighboring block.
In the method of decoding an image,

Extracting an AMVP index from the bitstream;

Selecting one or more representative blocks according to the priority of the motion compensation direction performed in each neighboring block among the neighboring blocks decoded using the inter prediction;

Identifying the extracted AMVP index in consideration of the selected representative block and candidate motion blocks of a predetermined neighboring picture; And

Generating a prediction block by reconstructing a motion vector of a block identified according to the AMVP index among the selected representative blocks and candidate motion blocks of the neighboring picture to a prediction motion vector of a current block, and reconstructing a motion vector of the current block

Image decoding method comprising a.
The method of claim 12,

In the step of selecting one or more representative blocks,

And setting a priority of the motion compensation direction for the past direction and the future direction, and selecting a block that performs the motion compensation in the direction having the highest priority among the neighboring blocks as the representative block.
The method of claim 12,

The representative block,

And selecting one from a left neighboring block group and an upper neighboring block group of the current block.
The method of claim 12, wherein in selecting the one or more representative blocks,

And selecting the representative block by further considering the priority of the distance to the picture referenced by each neighboring block.
The method of claim 15,

And the priority is determined based on whether the distance between the picture referenced by the current block and the picture referenced by the neighboring block is equal to each other.
In the method of decoding an image,

Extracting an AMVP index from the bitstream and selecting a representative block according to the priority of any one of an inter prediction method, an inter prediction direction, and a block size for a neighboring block decoded using inter prediction;

Identifying the extracted AMVP index in consideration of the selected representative block and candidate motion blocks of a predetermined neighboring picture; And

Generating a prediction block by reconstructing a motion vector of a block identified according to the AMVP index among the selected representative blocks and candidate motion blocks of the neighboring picture to a prediction motion vector of a current block, and reconstructing a motion vector of the current block

Image decoding method comprising a.