KR101316710B1 - Motion Vector Encoding/Decoding Method and Apparatus and Video Encoding/Decoding Method and Apparatus - Google Patents

Abstract

The present invention relates to a motion vector encoding method and apparatus, and an image encoding / decoding method and apparatus using the same.
An apparatus for encoding a motion vector, the present invention comprising: a prediction candidate selector for selecting one or more motion vector prediction candidates; A predicted motion vector determiner that determines a predicted motion vector among the one or more motion vector prediction candidates; And a difference vector encoder that calculates and encodes a difference vector by subtracting a predicted motion vector determined from a current motion vector.
According to the present invention, the compression efficiency can be improved by reducing the amount of bits for encoding the information on the predicted motion vector while encoding the motion vector using the predicted motion vector more accurately.

Description

Motion Vector Encoding Method and Apparatus and Image Encoding / Decoding Method and Apparatus Using Them {Motion Vector Encoding / Decoding Method and Apparatus and Video Encoding / Decoding Method and Apparatus}

The present invention relates to a motion vector coding method and apparatus, and a method and apparatus for image coding / decoding using the same. More specifically, inter prediction is performed by using a pixel that is encoded, decoded, and reconstructed before a block to be currently encoded, and efficiently encodes or decodes a motion vector used for inter prediction, and uses the data of a still image or a moving image. It relates to a method and apparatus for encoding or decoding.

As a method for efficiently compressing a video signal, there are JPEG, JPEG-2000, H.261, H.263, MPEG-2, MPEG-4, etc. for still images. In addition, MPEG-4 AVC (Advanced Video Coding) has been standardized in 2003 by the International Organization for Standardization (ISO), which provides further improved compression efficiency while inheriting technologies such as MPEG-2 and MPEG-4.

The encoding of data for a video is made up of intra prediction encoding and inter prediction encoding. Such intra prediction encoding or inter prediction encoding is widely used for compressing various data in an effective way to reduce correlation between data. In particular, since the motion vector of the current block determined by estimating the motion of the current block to be currently encoded by inter prediction encoding is closely correlated with the motion vector of the neighboring block, the motion vector of the neighboring block is determined from the motion vector of the current block. After calculating the predicted motion vector (PMV: Predicted Motion Vector), the differential value of the predicted value (DMV: Differential Motion Vector) is calculated without encoding the value of the motion vector of the current block itself. By coding only '), the amount of bits to be encoded can be significantly reduced, thereby improving the coding efficiency.

That is, in most existing image compression standards such as MPEG-4 AVC, when performing inter prediction encoding, the encoder estimates the motion of the current block in a previously encoded, decoded and reconstructed reference frame, and determines the current motion vector and the current. Only the difference vector, which is the difference with the motion vector, is encoded and transmitted. The decoder also predicts the motion vector of the current block by using the motion vector of the neighboring block previously decoded, and reconstructs the current motion vector by adding the transmitted difference vector and the predicted motion vector.

However, if the motion vector is encoded according to the existing image compression standards as described above, the difference vector increases when the predicted motion vector is not correlated with the current motion vector. In addition, even when the encoder additionally encodes information on the predictive motion vector in order to reduce the difference vector, there is a problem in that the bit amount due to encoding the additional information is increased to reduce the compression efficiency.

In order to solve the above-described problems, the present invention has a main purpose of improving a compression efficiency by reducing a bit amount for coding information on a predictive motion vector, while encoding a motion vector using a more precisely predicted motion vector .

According to an aspect of the present invention, there is provided an apparatus for encoding a current motion vector, which is a motion vector of a current block, comprising: a prediction candidate selector for selecting a motion vector prediction candidate for predicting the current motion vector; A predicted motion vector determiner that determines a predicted motion vector from the motion vector prediction candidate; And an encoder that calculates a difference vector by subtracting the prediction motion vector from the current motion vector, and encodes prediction candidate identification information and the difference vector for identifying the determined prediction motion vector, wherein the prediction candidate selector includes: A module for selecting a first motion vector prediction candidate from a neighboring block of the current block, and a module for selecting a second motion vector prediction candidate from a corresponding block at the same position as the current block or a neighboring block of the corresponding block in a reference frame It provides a motion vector encoding apparatus comprising a.

Another aspect of the present invention provides a method of encoding a current motion vector, which is a motion vector of a current block, comprising: selecting a motion vector prediction candidate for predicting the current motion vector; Determining a prediction motion vector from the motion vector prediction candidate; And calculating the difference vector by subtracting the prediction motion vector from the current motion vector, and encoding the prediction candidate identification information and the difference vector for identifying the determined prediction motion vector, and selecting the motion vector prediction candidate. The method may include selecting a first motion vector prediction candidate in a neighboring block of the current block, and selecting a second motion vector prediction candidate in a corresponding block at the same position as the current block or a neighboring block of the corresponding block in a reference frame. It provides a motion vector encoding method comprising the step of selecting.

According to still another aspect of the present invention, there is provided an apparatus for decoding a current motion vector of a current block, comprising: a decoder for reconstructing a difference vector and prediction candidate identification information from a bitstream; And selecting a motion vector prediction candidate for predicting the current motion vector, determining a motion vector prediction code identified by the prediction candidate identification information among the motion vector prediction candidates as a prediction motion vector, and determining the difference with the prediction motion vector. A motion vector decompressor for adding a vector to reconstruct a current motion vector of the current block, wherein the motion vector decompressor comprises: a module for selecting a first motion vector prediction candidate from a neighboring block of the current block; And a module for selecting a second motion vector prediction candidate from a corresponding block located at the same position as the block or a neighboring block of the corresponding block.

Another aspect of the present invention is a method of decoding a current motion vector of a current block, comprising: reconstructing a difference vector from a bitstream; Selecting a motion vector prediction candidate for predicting the current motion vector; Restoring prediction candidate identification information from the bitstream; Determining a motion vector prediction candidate identified by the prediction candidate identification information among the motion vector prediction candidates as a prediction motion vector; And reconstructing the current motion vector of the current block by adding the prediction motion vector and the difference vector, wherein the selecting of the motion vector prediction candidate comprises selecting a first motion vector prediction candidate from a neighboring block of the current block. And selecting a second motion vector prediction candidate from a corresponding block at the same position as the current block or a neighboring block of the corresponding block in a reference frame.

As described above, according to the present invention, the compression efficiency can be improved by reducing the amount of bits for encoding information on the predicted motion vector while encoding the motion vector using the predicted motion vector more accurately.

1 is a block diagram schematically illustrating an image encoding apparatus according to an embodiment of the present invention.
2 is a block diagram schematically showing a motion vector coding apparatus according to an embodiment of the present invention.
3 is an exemplary diagram showing a zigzag scan direction;
4 is an exemplary diagram illustrating a neighboring block of a current block;
5 is an exemplary diagram illustrating a current block and a neighboring block of a current frame;
6 is an exemplary diagram illustrating blocks of a current frame and a reference frame by way of example;
7 is an exemplary diagram for explaining an instruction vector;
8 is an exemplary diagram for explaining a process of calculating a threshold value;
9 is an exemplary diagram illustrating a motion vector prediction candidate and a current motion vector separated for each component;
10 is an exemplary diagram illustrating a prediction candidate identification flag according to a plurality of thresholds;
11 shows an example of transmission bits of a prediction candidate identification flag;
12 is a flowchart illustrating a motion vector encoding method according to an embodiment of the present invention;
13 is a flowchart illustrating a video encoding method according to an embodiment of the present invention;
14 is a block diagram schematically illustrating an image decoding apparatus according to an embodiment of the present invention;
15 is a block diagram schematically illustrating a motion vector decoding apparatus according to an embodiment of the present invention;
16 is a flowchart for explaining a motion vector decoding method according to an embodiment of the present invention;
17 is a flowchart illustrating an image decoding method according to an embodiment of the present invention.

Hereinafter, some embodiments of the present invention will be described in detail with reference to exemplary drawings. In adding reference numerals to the components of each drawing, it should be noted that the same reference numerals are assigned to the same components as much as possible even though they are shown in different drawings. In the following description of the present invention, a detailed description of known functions and configurations incorporated herein will be omitted when it may make the subject matter of the present invention rather unclear.

In addition, in describing the component of this invention, terms, such as 1st, 2nd, A, B, (a), (b), can be used. These terms are intended to distinguish the constituent elements from other constituent elements, and the terms do not limit the nature, order or order of the constituent elements. When a component is described as being "connected", "coupled", or "connected" to another component, the component may be directly connected or connected to the other component, Quot; may be "connected," "coupled," or "connected. &Quot;

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

The image encoding apparatus 100 according to an embodiment of the present invention may include a predictor 110, a subtractor 120, a transformer 130, a quantizer 140, an encoder 150, an inverse quantizer 160, and an inverse converter. 170, the adder 180, and the memory 190 may be configured. Here, the inverse quantizer 160, the inverse transformer 170, the adder 180, and the memory 190 may be selectively included in the image encoding apparatus 100. The video encoding apparatus 100 may be a personal computer (PC), a notebook computer, a personal digital assistant (PDA), a portable multimedia player (PMP), or a PlayStation Portable (PSP). ), 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, and executing a program. Means a variety of devices including a microprocessor for operation and control.

The predictor 110 determines a current motion vector, which is a motion vector of the current block, and generates a prediction block using the determined current motion vector. That is, the predictor 110 predicts a pixel value of each pixel of the current block to be encoded in the image by using a current motion vector, and has a predicted pixel value of the predicted pixel value of each pixel predicted. Create a Predicted Block. The predictor 110 may determine a current motion vector by using various techniques such as rate-distortion optimization (RDO), and the determined current motion vector may be an optimal motion vector.

The subtractor 120 subtracts the current block and the prediction block to generate a residual block. That is, the subtractor 120 calculates a difference between the pixel value of each pixel of the current block to be encoded and the predicted pixel value of each pixel of the prediction block predicted by the predictor 110 to obtain a residual signal in the form of a block. Create a residual block with

Transformer 130 transforms the residual block. That is, the converter 130 converts the residual signal of the residual block output from the subtractor 120 into the frequency domain to convert each pixel value of the residual block into a frequency coefficient. Here, the transformer 130 uses residuals using various transformation techniques for transforming an image signal of a spatial axis into a frequency axis such as a Hadamard transform and a Discrete Cosine Transform Based Transform (DCT based Transform). The signal can be converted into the frequency domain, and the residual signal converted into the frequency domain becomes a frequency coefficient.

Quantizer 140 quantizes the transformed residual block. That is, the quantizer 140 quantizes the frequency coefficients of the residual block output from the transformer 130 and outputs the residual block having the quantized frequency coefficients. Here, the quantizer 140 may be quantized using Dead Zone Uniform Threshold Quantization (DZUTQ), a quantization weighted matrix, or an improved quantization technique. Can be.

In the above description, the image encoding apparatus 100 according to the exemplary embodiment of the present invention includes and illustrates a transformer 130 and a quantizer 140. However, the transformer 130 and the quantizer 140 are optional. It may be included as. That is, the image encoding apparatus 100 according to an embodiment of the present invention may generate a residual block having a frequency coefficient by converting the residual signal of the residual block and do not perform a quantization process, and the residual signal of the residual block may be frequency Not only can we perform the quantization process without converting to coefficients, we can even not perform both the transformation and quantization processes.

The encoder 150 encodes the quantized residual block, determines a predicted motion vector from one or more motion vector prediction candidates, and subtracts the predicted motion vector from the current motion vector. (Differential Motion Vector) is encoded, and a bitstream including an encoded residual block and an encoded difference vector is generated and output. That is, the encoder 150 scans the quantization frequency coefficients of the residual block output from the quantizer 140 according to various scan methods such as a zigzag scan illustrated in FIG. 3 to generate a quantization frequency coefficient string, and entropy encoding ( It is encoded using various encoding techniques such as Entropy Coding).

In addition, the encoder 150 encodes a current motion vector, selects one or more motion vector prediction candidates, and determines a prediction motion vector among the selected one or more motion vector prediction candidates. To this end, the encoder 150 may include a motion vector of a neighboring block of the current block, a motion vector of a Collocated Block (Col) block at the same position as the current block in the reference frame, a motion vector of a neighboring block of the Col block, and a current block. At least one of the indication vectors indicating a specific block of the reference frame may be selected as one or more motion vector prediction candidates. In this case, the specific block may be a block having a motion vector that is horizontal to the indication vector or smaller than a predetermined threshold.

For example, when the encoder 150 determines a motion vector prediction candidate of one of the one or more motion vector prediction candidates as a prediction motion vector, as illustrated in FIG. 4, the encoder 150 may determine the prediction motion vector using a median value. Can be. Referring to FIG. 4, assuming that the current block is X, the left block of the current block is A, the top block of the current block is B, and the top right block of the current block is C, the motion vector of the left block A is MV_A. When the motion vector of the upper block B is MV_B and the motion vector of the upper right block C is MV_C, the predicted motion vector of the current block A may be calculated using an intermediate value, as shown in Equation 1 below.

The encoder 150 may determine the prediction motion vector from among one or more motion vector prediction candidates using the median value, but it is not necessary to use the median value and may determine the prediction motion vector using various other methods. This will be described in detail with reference to FIG. 2 in the following process.

In addition, the encoder 150 may generate and encode a prediction candidate identification flag that identifies a motion vector prediction candidate determined as the prediction motion vector, and then include the same in the bitstream. That is, the bitstream may include an encoded prediction candidate identification flag as well as an encoded residual block and an encoded difference vector.

In addition, the encoder 150 may independently determine each of the x-component and the y-component of the prediction motion vector. That is, the encoder 150 determines the x-component of one motion vector prediction candidate among the x-components of the one or more motion vector prediction candidates as the x-component of the prediction motion vector, and the y-component of the one or more motion vector prediction candidates. The y-component of one motion vector prediction candidate may be determined as the y-component of the predictive motion vector.

In addition, the encoder 150 may determine that at least one of the absolute value of the x-component of the difference vector and the absolute value of the y-component of the difference vector is greater than one or more of the preset x-component threshold and the preset y-component threshold, respectively. In this case, the prediction candidate identification flag identifying the motion vector prediction candidate determined as at least one of the x-component of the prediction motion vector and the y-component of the prediction motion vector may be generated, encoded, and further included in the bitstream. . The encoder 150 will be described in detail with reference to FIG. 2 in a later process.

Inverse quantizer 160 inverse quantizes the residual block quantized by quantizer 140. The inverse transformer 170 inverse transforms a residual block inversely quantized by the inverse quantizer 160. Here, the inverse quantizer 160 and the inverse transformer 170 may perform inverse quantization and inverse transformation by using the quantization method and the transformation method used in the quantizer 140 and the transformer 130 inversely. In addition, when only the quantization is performed in the transformer 130 and the inverse quantizer 140 and no transformation is performed, only the inverse quantization may be performed and the inverse transformation may not be performed. If both transformation and quantization are not performed, the inverse quantizer 160 and the inverse transformer 170 may also be omitted without being included in the inverse transformation and inverse quantization or included in the image encoding apparatus 100.

The adder 180 reconstructs the current block by adding the prediction block predicted by the predictor 110 and the residual block reconstructed by the inverse transformer 170. The memory 190 stores the reconstructed current block output from the adder 180 as a reference frame in units of frames so that the predictor 110 can use it as a reference frame when encoding the next block of the current block or another block in the future. .

Although not shown in FIG. 1, the image encoding apparatus 100 according to an embodiment of the present invention may include an intra predictor for intra prediction and a reconstructed current block based on the H.264 / AVC standard. The deblocking filter may further include a deblocking filtering. Here, deblocking filtering refers to a task of reducing block distortion caused by coding an image block by block. The deblocking filter may be applied to a block boundary or a macro block boundary, a deblocking filter may be applied to a macro block boundary, Can be selectively used.

2 is a block diagram schematically illustrating a motion vector coding apparatus according to an embodiment of the present invention.

The motion vector encoding apparatus according to an embodiment of the present invention may be implemented with the encoder 150 of the image encoding apparatus 100 according to the embodiment of the present invention described above with reference to FIG. 1. Hereinafter, for convenience of description, the motion vector encoding apparatus according to an embodiment of the present invention is called an encoder 150. In addition, the encoder 150 according to an embodiment of the present invention may further include an image encoder for outputting the encoded residual block by encoding the quantized residual block, as described above with reference to FIG. 1. Since the encoder is obvious to those skilled in the art, detailed description thereof will be omitted.

The encoder 150 according to an embodiment of the present invention includes a prediction candidate selector 210, a prediction motion vector determiner 220, a differential vector encoder 230, a prediction candidate identification flag encoder 240, and a prediction candidate selection flag encoder ( 250). Here, the encoder 150 may selectively include one or more of the prediction candidate identification flag encoder 240 and the prediction candidate selection flag encoder 250 without necessarily including them.

The prediction candidate selector 210 selects one or more motion vector prediction candidates. The predictive motion vector determiner 220 determines one motion vector prediction candidate among the one or more motion vector prediction candidates as the prediction motion vector. The difference vector encoder 230 calculates and encodes a difference vector by subtracting a prediction motion vector from the current motion vector. The prediction candidate identification flag encoder 240 generates and encodes a prediction candidate identification flag that identifies a motion vector prediction candidate determined as the prediction motion vector. Prediction candidate selection flag encoder 250 is a prediction candidate for identifying which motion vector prediction candidate is selected by prediction candidate selector 210 for every region of the image (for example, a slice in H.264). Generate and encode a selection flag.

The difference vector may be calculated as in Equation 2. In Equation 2, MVd represents a difference vector, MV represents a current motion vector, and MVp_opt represents a predicted motion vector.

The difference vector encoded by the difference vector encoder 230 and the encoded residual block output from the image encoder described above may be included in a bitstream and transmitted to the image decoding apparatus. In addition, the bitstream may further include one or more of the prediction candidate selection flag encoded by the prediction candidate selection flag encoder 250 and the prediction candidate identification flag encoded by the prediction candidate identification flag encoder 240.

Here, the prediction candidate selector 210 may select a motion vector of a neighboring block of the current block as one or more motion vector prediction candidates. That is, referring to FIG. 5 exemplarily illustrating a current block and a neighboring block of a current frame, one or more motion vector prediction candidates are neighboring blocks located around a current block X in a current frame where a current block X to be currently encoded is located. It may be motion vectors of L, UL, U, and UR. In FIG. 5, only the left block L, the top left block UL, the top block U, and the top right block UR of the current block are shown as neighboring blocks, but the motion vector prediction candidate is not only the other block adjacent to the current block but also the movement of other non-adjacent blocks. Can be vectors.

In addition, the prediction candidate selector 210 may select one or more of the motion vector of the Col block and the motion vector of the neighboring block of the Col block at the same position as the current block in the reference frame as one or more motion vector prediction candidates. That is, referring to FIG. 6 exemplarily showing blocks of the current frame and the reference frame, one or more motion vector prediction candidates are included in the Col block and the Col block, which are blocks located at the same position as the current block X in the reference frame frame t-1. It may be motion vectors of neighboring blocks (UL _Col , U _Col , UR _Col , L _Col , R _Col , BL _Col , B _Col , BR _Col ). In FIG. 6, only the Col block and the neighboring blocks (UL _Col , U _Col , UR _Col , L _Col , R _Col , BL _Col , B _Col , BR _Col ) of the Col block are shown, but not only the motion vector prediction candidate is a Col block. And other blocks adjacent to, as well as motion vectors of other non-adjacent blocks.

In addition, the prediction candidate selector 210 may select, as one or more motion vector prediction candidates, an indication vector indicating a specific block of the reference frame in the current block of the current frame. That is, referring to FIG. 7 illustrated by way of example to describe the indication vector, a vector indicating a first par frame and a vector indicating any block Par located in a first reference frame frame t-1 in a current block X located in a current frame frame t If the motion vectors of any of the blocks Par located in frame t-1 are horizontal to each other or the difference is smaller than the preset threshold, a vector indicating any block Par of the reference picture frame t-1 in the current block X of the current frame frame t It may be selected as a motion vector prediction candidate. Here, the vector indicating any block Par of the reference picture frame t-1 at the position of the current block X of the current frame f is an indicator vector. Therefore, the specific block described above may be a block having a motion vector that is horizontal to the instruction vector or smaller than a predetermined threshold.

When the prediction candidate selector 210 selects one or more motion vectors, the prediction candidate selection flag encoder 250 generates and encodes a prediction candidate selection flag for identifying the motion vector prediction candidate selected by the prediction candidate selector 210. The encoded prediction candidate selection flag may be inserted in the header of the bitstream.

As another example, the prediction candidate selector 210 may select some motion vector prediction candidates from one or more motion vector prediction candidates. That is, the prediction candidate selector 210 may select several of the one or more motion vector prediction candidates according to a predetermined criterion according to the characteristics of the image to determine the prediction motion vector only among the selected motion vector prediction candidates. In this case, the prediction candidate selection flag encoder 250 may determine a portion of the one selected by the prediction candidate selector 210 to identify which one of the one or more motion vector prediction candidates is selected as some motion vector prediction candidate. A prediction candidate selection flag for identifying a motion vector prediction candidate may be generated and encoded.

Here, the predetermined criterion may be a probability that each of the one or more motion vector prediction candidates may be determined as the current motion vector. When the probability is used, the prediction candidate selector 210 uses each of the one or more motion vector prediction candidates. By encoding one or more regions of the current frame, a probability determined as the current motion vector may be calculated for each of the one or more motion vector prediction candidates, and some motion vector prediction candidates may be selected according to the calculated probability. For example, assuming that one or more motion vector prediction candidates total 5 A, B, C, D, and E, each of the 5 motion vector prediction candidates is virtually encoded to optimally encode arbitrary regions of the current frame. Calculate the probability to be determined as the motion vector of (i.e., the probability to be determined as the current motion vector), and if the calculated result is A: 80%, B: 10%, C: 5%, D: 3%, E: 2% If calculated as a probability, A or A and B motion vector prediction candidates may be selected as some motion vector prediction candidates.

In addition, the predetermined criterion may be correlation. When the correlation is used, the prediction candidate selector 210 selects correlated motion vector prediction candidates from one or more motion vector prediction candidates into one or more groups, and selects each selected group. Representative representative motion vectors can be selected as some motion vector prediction candidates. Here, the representative motion vector may be an intermediate value or an average value of motion vector prediction candidates in each selected group.

For example, if the one or more motion vector prediction candidates are the motion vectors described above with reference to FIGS. 5 to 7, the neighboring blocks L, UL, U, and UR of some of the neighboring blocks of the current block described above with reference to FIG. Collect only the motion vectors MV_L, MV_UL, MV_U, MV_UR, and select one group A. Determine the value MVp_Group A calculated through the intermediate value calculation as a representative motion vector and partially represent the representative motion vector as shown in Equation 3. Can be selected as a motion vector prediction candidate.

In addition, the motion vectors MV_Col, MV_U _col , MV_L _col , MV_R of the top, bottom, left, and right blocks Col, U _Col , L _Col , R _Col , and B _Col among the Col block and the neighboring blocks of the Col block described with reference to FIG. 6. _col , MV_B _col ) can be collected as one group B, and as shown in Equation 4, the value MVp_Group A calculated through the average value operation can be determined as a representative motion vector, and the representative motion vector can be selected as a part of motion vector prediction candidate. have.

Also, the motion vectors MV_Col and MV_UL _{col of} the Col block and the neighboring blocks (UL _Col , U _Col , UR _Col , L _Col , R _Col , BL _Col , B _Col , BR _Col ) of the Col block described above with reference to FIG. 6. , MV_U _col , MV_UR _col , MV_L _col , MV_R _col , MV_BL _col , MV_B _col , MV_BR _col ), and select them as one group C, and represent the value MVp_Group C calculated through the median operation, as shown in Equation 5. It can be determined as a motion vector and a representative motion vector can be selected as some motion vector prediction candidate.

As described above, the representative motion vector calculated by the equations (4) to (5) and the instruction vector described above with reference to FIG. 7 may be selected as some motion vector prediction candidates. In this case, the prediction candidate selection flag encoder ( 250 may generate and encode a prediction candidate selection flag to indicate that MVp_groupA, MVp_groupB, MVp_groupC, MV_col and MV_par are selected as some motion vector prediction candidates. Of course, the prediction candidate identification flag encoder 240 generates and encodes a prediction candidate identification flag for indicating which motion vector prediction candidates among some motion vector prediction candidates MVp_groupA, MVp_groupB, MVp_groupC, MV_col and MV_par have been determined as the prediction motion vectors. can do.

Also, the prediction candidate selector 210 may select one or more motion vector prediction candidates, which are predetermined from among the one or more motion vector prediction candidates, as partial motion vector prediction candidates. Here, the one or more preset motion vector prediction candidates may be arbitrarily selected by the user.

In addition, the predictive motion vector determiner 220 may determine the predictive motion vector in various ways. As an example, the prediction motion vector determiner 220 may use the minimum amount of bits required to encode a difference vector calculated using each motion vector prediction candidate among one or more motion vector prediction candidates selected by the prediction candidate selector 210. A motion vector prediction candidate to be determined may be determined as a prediction motion vector. Accordingly, the prediction motion vector determiner 220 calculates a difference vector by assuming that each motion vector prediction candidate is a prediction motion vector, encodes the calculated difference vector, and selects a motion vector prediction candidate whose encoded data is minimum as a prediction motion vector. You can decide. In this case, the predictive candidate identification flag encoder 240 encodes a motion vector of the current block (that is, when encoding a difference vector between the current motion vector and the determined predicted motion vector). A prediction candidate identification flag that identifies a motion vector prediction candidate determined as the prediction motion vector may be generated and encoded to identify whether the vector prediction candidate has been determined as the prediction motion vector.

As another example, the predictive motion vector determiner 220 may independently determine each of the x-component and the y-component of the predictive motion vector. That is, the prediction motion vector determiner 220 determines an x-component of one motion vector prediction candidate among the x-components of the one or more motion vector prediction candidates as an x-component of the prediction motion vector, The y-component of one motion vector prediction candidate among the y-component may be determined as the y-component of the predictive motion vector. For example, in encoding a difference vector, the amount of bits of the difference vector is proportional to the absolute value of the magnitude of the difference vector, and one or more motion vector prediction candidates are selected from (-4, -4), (2, 2), (4, 4), (14, 14) and the current motion vector is (0, 4), conventionally (2, 2) or (4, 4) can be determined as the predicted motion vector, but also in the above-described example The x-component of the motion vector may be selected from 2 out of -4, 2, 4 and 14, and the y-component of the predictive motion vector may be selected from 4 out of -4, 2, 4 and 14. In other words, the x-component and the y-component cannot be separated and selected as in the prior art, but the optimum value can be determined as the predicted motion vector by separating the x-component and the y-component.

In this case, the predictive candidate identification flag encoder 240 sets the absolute value of the x-component of the difference vector and the absolute value of the y-component of the difference vector, respectively, to the preset x-component threshold and the preset y-component threshold. In the same case, the prediction candidate identification flag may not be generated and thus the prediction candidate identification flag may not be encoded and transmitted. In addition, the prediction candidate identification flag encoder 240 may set at least one of an absolute value of the x-component of the difference vector and an absolute value of the y-component of the difference vector, respectively, to a preset x-component threshold and a preset y-component threshold. If less than or equal to one of the following, a prediction candidate identification flag for identifying a motion vector prediction candidate determined as at least one of an x-component of the predictive motion vector and a y-component of the predictive motion vector may be generated and encoded. That is, if two of the two components (x-component and y-component) of the prediction motion vector are larger than the threshold, the prediction candidate identification flag is not generated. If only one component is smaller than the threshold, the threshold is generated. Generate a prediction candidate identification flag that identifies which motion vector prediction candidate was used as one larger component, and if two components are not greater than the threshold, determine which motion vector prediction candidate was used by the two components. A predictive candidate identification flag to be identified can be generated and encoded.

Here, when the absolute value of each component of the difference vector is greater than the threshold value of each component, the prediction candidate identification flag is not generated and encoded, and the prediction candidate is less than or equal to the threshold value of each component. When generating and encoding the identification flag, a prediction candidate identification flag for identifying which motion vector prediction candidate among the motion vector prediction candidates selected for each is determined as the prediction motion vector may be separately generated and encoded. In the image decoding apparatus, the prediction candidate identification flag is not transmitted because the absolute value of each component of the difference vector is larger than the corresponding component threshold and the absolute value of each component of the difference vector is less than or equal to the corresponding component threshold. For the case where is transmitted, the motion vector prediction candidates may be separately decoded to determine the prediction motion vector.

For example, if one of the selected motion vector prediction candidates of the group A is determined as a predicted motion vector to calculate a difference vector, and if the absolute value of each component of the difference vector is greater than a threshold value of the corresponding component, If the absolute value of each component of the difference vector is smaller than the threshold value of the corresponding component, one of the selected motion vector prediction candidates of the group B is determined as a predicted motion vector to calculate a difference vector, A prediction candidate identification flag that identifies which of the selected motion vector prediction candidates of B has been determined as a prediction motion vector is generated and encoded. That is, when the absolute value of each component of the difference vector is greater than the threshold of the component, the selected motion vector prediction candidates of group A are used, and the absolute value of each component of the difference vector is less than or equal to the threshold of the corresponding component. In this case, the selected motion vector prediction candidates of group B may be used.

Here, each component threshold value (x-component threshold value and y-component threshold value) can be calculated in various ways. For example, the x-component threshold may be the maximum value of the difference between the x-components of the two closest motion vector prediction candidates among the x-components of the one or more motion vector prediction candidates divided by a predetermined number, and the y-component. The threshold may be a maximum value of values obtained by dividing a difference value between y-components of two nearest motion vector prediction candidates among y-components of one or more motion vector prediction candidates by a predetermined number. That is, each component is the MVp _Cand (i) and the two nearest motion vector prediction candidates as shown in Equation (6 ) . After calculating thre _ij for each MVp _Cand (j) ), it may be calculated as the maximum value of the calculated thre _ij .

Referring to FIG. 8, which is illustrated by way of example to explain a process of calculating a threshold value, as shown in FIG. 8, the value of the x-component of the motion vector prediction candidate is '-4', '2', and '4'. , Assuming '14', the value obtained by dividing the difference of the x-component value of each motion vector prediction candidate by 2 is '3' in interval ①, '1' in interval ②, and '5' in interval ③. Since the maximum value is '5', the threshold value is '5'.

Referring to FIG. 9 in which a motion vector prediction candidate and a current motion vector are separately illustrated for each component, the threshold of the x-component is '2' and the threshold of the y-component is '3'. In this case, the x-component of the predictive motion vector is determined to be '3' (MVp_cand [2]) so that the x-component of the differential vector is '-3' and the y-component of the predictive motion vector is '2' (MVp_cand [0]), and the y-component of the difference vector becomes '+4'. Since the absolute values of the two components of the difference vector are larger than the thresholds of each component, the prediction candidate identification flag encoder 240 predicts which motion vector prediction candidates among the motion vector prediction candidates are determined as the prediction motion vectors. An identification flag may not be generated and encoded. As described above, in the image decoding apparatus, when reconstructing the current motion vector, even if it is not known which motion vector prediction candidate is determined as the prediction motion vector among the one or more motion vector prediction candidates because there is no prediction candidate identification flag, the difference vector is determined. If x-component and y-component of are greater than the threshold value of each component, and if it is larger than the threshold value, it is possible to determine which motion vector prediction candidate was determined as the prediction motion vector by itself.

Also, as another example, the x-component thresholds may be values obtained by dividing a difference value between x-components of two closest motion vector prediction candidates among x-components of one or more motion vector prediction candidates by a predetermined number, and y-component. The thresholds may be values obtained by dividing the difference between the y-components of the two closest motion vector prediction candidates among the y-components of the one or more motion vector prediction candidates by a predetermined number. That is, each component threshold value is not the largest value among the distances between the motion vector prediction candidates divided by 2 as shown in Equation 6, but the values obtained by dividing the distance between the motion vector prediction candidates by 2 as shown in Equation 7 below. It can be calculated as Therefore, when there are N motion vector prediction candidates, the threshold value of each component may be N-1. However, in Equation 7, 0 ≦ n <N.

For example, Figure 8, and a case, the first threshold value of the x- component (Threshold ₁₎ may be a "3", the second threshold (Threshold ₂₎ may be a "1", the third The threshold ₃ may be '5'.

In this case, the prediction candidate identification flag encoder 240 may generate and encode the prediction candidate identification flag differently according to the magnitude of the absolute value of each component of the difference vector and the threshold of each component. That is, the prediction candidate identification flag encoder 240 sorts the N-1 thresholds in descending order so that when the absolute value of each component of the difference vector is less than or equal to the smallest threshold value of the corresponding component, the N selected motion vectors are selected. Generate and encode a prediction candidate identification flag that identifies which motion vector prediction candidate is determined as the prediction motion vector among the prediction candidates, wherein the absolute value of each component of the difference vector is less than or equal to the second smallest threshold of that component. In this case, it is assumed that there is no current motion vector between two motion vector prediction candidates having the smallest threshold value of the corresponding component, and which motion vector prediction candidate is determined as the prediction motion vector among the N-1 selected motion vector prediction candidates. By generating a prediction candidate identification flag that identifies the Encode If the absolute value of each component of the difference vector is less than or equal to the third smallest threshold of the component, the assumption is that there is no current motion vector between the two motion vector prediction candidates with the smallest threshold of the component. Assuming that there is no current motion vector between the two motion vector prediction candidates having the second smallest threshold of the component, a motion vector prediction candidate is determined as the prediction motion vector among the N-2 selected motion vector prediction candidates. And generate a prediction candidate identification flag to identify whether or not.

In this way, a predictive candidate identification flag is generated and encoded by comparing the absolute value of each component of the difference vector with each threshold value of the corresponding component, and the absolute value of each component of the difference vector is greater than all the threshold values of the corresponding component. In this case, it is not necessary to generate and encode a prediction candidate identification flag for identifying which of the N motion vector prediction candidates is determined as the prediction motion vector. As such, depending on the absolute value of each component of the difference vector and the threshold values of the corresponding component, the length of the prediction candidate identification flag for identifying which motion vector prediction candidate is determined as the prediction motion vector may vary.

According to the above method, when the absolute value of each component of the difference vector is larger than all the thresholds, the prediction motion vector may be determined by individually selecting each component of the motion vector prediction candidate. If only one component of the absolute value of each component of the difference vector is greater than all the thresholds, the motion vector prediction candidates are predicted among the motion vector prediction candidates only for the components of the difference vector having an absolute value not greater than all the thresholds. A prediction candidate identification flag is sent to identify whether the decision was made as a vector.

That is, when only one of the absolute values of two components of the difference vector exceeds the largest threshold, a prediction candidate identification flag for identifying a motion vector prediction candidate having a component having an absolute value not exceeding the largest threshold is generated. Can be encoded. If the absolute values of both components of the difference vector are not greater than the largest threshold, a prediction candidate identification flag is generated and encoded to independently identify for each component which motion vector prediction candidate is determined as the prediction motion vector. Rather, two components may simultaneously generate and encode a prediction candidate identification flag for identifying a motion vector prediction candidate determined as a prediction motion vector.

For example, referring to FIG. 8 and FIG. 10 exemplarily illustrating a prediction candidate identification flag according to a plurality of thresholds, the number of prediction candidates is four and each threshold is '1', '3', Becomes '5'. Accordingly, FIG. 10 illustrates a section divided according to three threshold values. The interval denoted by ① indicates the x-components of the motion vector prediction candidates, and the interval denoted by ② denotes the current motion vector when the absolute value of the x-component of the difference vector is less than or equal to the smallest threshold of the x-component. where the x-component is located, and the section marked with ③ is the x-component of the current motion vector when the absolute value of the x-component of the difference vector is less than or equal to the second smallest threshold of the x-component. The interval indicated by ④ indicates the interval in which the x-component of the current motion vector is located when the absolute value of the x-component of the difference vector is less than or equal to the largest threshold of the x-component. , The interval ⑤ denotes the x-component of the current motion vector when the absolute value of the x-component of the difference vector is greater than the largest threshold of the x-component. A shows a section in which location.

The prediction candidate identification flag encoder 240 may generate and encode a prediction candidate identification flag that identifies which motion vector prediction candidate is determined as the prediction motion vector among the four motion vector prediction candidates in the section denoted by ②. In the marked section, a prediction candidate identification flag for identifying which motion vector prediction candidate is determined as the prediction motion vector among three motion vector prediction candidates may be generated and encoded. In the section indicated by ④, any of the two motion vector prediction candidates may be generated. A prediction candidate identification flag for identifying whether the motion vector prediction candidate is determined as the prediction motion vector may be generated and encoded, and the prediction candidate identification flag may not be generated in the section denoted by 5.

When the x-component of the current motion vector is '8', the x-component of the predictive motion vector is '4'. At this time, since the x-component of the difference vector is '+4', its absolute value is smaller than '5', which is the largest threshold of the x-component, and is greater than '3', which is the second largest threshold, so that the motion vector is predicted. The prediction candidate identification flag for identifying one of the x-components '4' and '14' of the candidate may be generated and encoded.

Since the x-component of the difference vector is '+4' in the image decoding apparatus, when the x-component '-4' of the motion vector prediction candidate is assumed to be the x-component of the predictive motion vector, the x-component of the current motion vector is determined. Can be assumed to be '0'. However, assuming that the x-component of the current motion vector is '0', the x-component of the optimal (i.e., minimizing the value of the x-component of the differential vector) predicted motion vector becomes '2', so the motion It can be seen that the x-component '-4' of the vector prediction candidate is not the prediction motion vector. In this way, it can be seen that the x-components '-4' and '2' of the motion vector prediction candidate are not determined as the prediction motion vectors. Accordingly, by generating and encoding a prediction candidate identification flag for identifying which value among the x-components '4' and '14' of the motion vector prediction candidate that can be determined as the prediction motion vector is determined as the x-component of the prediction motion vector Just do it. As described above, when a plurality of thresholds are calculated for each component, and the prediction candidate identification flag is generated and encoded differently by comparing the absolute values of each component of the difference vector and the plurality of threshold values of the corresponding component, the current motion vector is encoded. In this case, the coded bit amount of the prediction candidate identification flag to be transmitted can be reduced, thereby improving the compression efficiency.

The image decoding apparatus may identify which motion vector prediction candidate is determined as the prediction motion vector using only the prediction candidate identification flag.

As another example, the predicted motion vector determiner 220 selects only one of two motion vector prediction candidates smaller than any threshold value when the distance between each motion vector prediction candidate is smaller than a certain threshold among the selected motion vector prediction candidates. A motion vector prediction candidate of the remaining motion vector prediction candidates not excluded may be determined as the prediction motion vector except for the other motion vector prediction candidates as motion vector prediction candidates as regarded as selected. In the video decoding apparatus, when the distance between the motion vector prediction candidates among the selected motion vector prediction candidates is smaller than an arbitrary threshold value, only one of two motion vector prediction candidates smaller than an arbitrary threshold value is considered to be selected, and the other motion vector prediction candidates are regarded as motion vector. By excluding from the prediction candidates, one of the remaining motion vector prediction candidates can be determined as the prediction motion vector.

For example, when an arbitrary threshold value is '2' and three motion vector prediction candidates are selected, the first motion vector prediction candidate is '-5', the second motion vector prediction candidate is '8' If the motion vector prediction candidate is '9', since the distance between the second motion vector prediction candidate and the third motion vector prediction candidate is smaller than a certain threshold value, the third motion vector prediction candidate is excluded from the motion vector prediction candidates, It is possible to determine one of the motion vector prediction candidate and the second motion vector prediction candidate as a prediction motion vector. Using this method, the number of bits for encoding the prediction candidate identification flag can be reduced by reducing the number of selected motion vector prediction candidates.

In addition, when generating the prediction candidate identification flag, the prediction candidate identification flag encoder 240 may generate the prediction candidate identification flag in various ways. As an example, when the motion vector prediction candidate determined as the prediction motion vector is the motion vector prediction candidate having the highest probability of being determined as the current motion vector among the one or more motion vector prediction candidates, the prediction candidate identification flag encoder 240 may determine that the prediction candidate identification flag is the same. The prediction candidate identification flag may be generated to be the minimum bit.

That is, as shown in Equation 8, the maximum length L necessary for encoding the prediction candidate identification flag is calculated using the number NumOfMVp _Cand of the motion vector prediction candidates.

As shown in Equation 9, when the value of M obtained by subtracting '1' from the number of calculated motion vector prediction candidates and taking log ₂ is an integer, as shown in FIG. 11 exemplarily illustrating the transmission bit of the prediction candidate identification flag, the prediction motion When generating and encoding a prediction candidate identification flag for identifying a motion vector prediction candidate determined as a vector, when the most likely motion vector prediction candidate is a prediction motion vector, the prediction candidate identification flag is generated and encoded by one bit.

For example, if there are three selected motion vector prediction candidates (MVp_cand [i]: 0 ≦ i <3), since the maximum bit length L required for encoding is '2', any motion vector prediction candidate among the selected motion vector prediction candidates is selected. A 2-bit prediction candidate identification flag may be generated that identifies whether is determined as a predicted motion vector. At this time, since the value of M is an integer according to Equation 9, the motion vector prediction candidate having the highest probability is generated and encoded with one bit. If one bit of the prediction candidate identification flag is '1', the motion vector prediction candidate having the highest probability is determined as the prediction motion vector, and if it is '0', the motion vector prediction candidate having the highest probability is the prediction motion vector. Since it is not determined as, the second bit is generated and encoded by generating and encoding additional bits, thereby generating and encoding two bits for identifying the remaining two motion vector prediction candidates. If M is an integer, the image decoding apparatus reads only one bit of the prediction candidate identification flag to identify whether the motion vector prediction candidate having the highest probability is determined as the prediction motion vector, and the motion vector prediction candidate having the highest probability is determined as the prediction motion vector. Only if not, read the bit further to identify which motion vector prediction candidate was determined as the predicted motion vector.

In this case, the prediction candidate identification flag encoder 240 may calculate the probability to be determined as the current motion vector for each of the one or more motion vector prediction candidates by encoding one or more regions of the current frame using the one or more motion vector prediction candidates. A motion vector prediction candidate that is randomly designated or set according to a randomly designated probability may be selected as a motion vector prediction candidate having the highest probability.

As another example, the prediction candidate identification flag encoder 240 may generate and encode the prediction candidate identification flag only when at least one or more of the one or more motion vector prediction candidates are not the same. That is, the prediction candidates selected by the selected motion vector prediction candidates are regarded as one. If all the selected prediction candidates are all the same, the corresponding motion vector prediction candidate is optimal even if any motion vector prediction candidate is determined as the prediction motion vector. Therefore, it is identified which motion vector prediction candidate is determined as the prediction motion vector among the motion vector prediction candidates. The prediction candidate identification flag may be generated and not encoded. In this case, the video decoding apparatus does not decode the prediction candidate identification flag.

As another example, the prediction candidate identification flag encoder 240 generates a prediction candidate identification flag for each case when the block type of the current block is the SKIP mode, the P block, or the B block. Can be encoded. That is, the prediction motion vector determiner 220 may select different motion vector prediction candidates for each of the block type of the current block in the SKIP mode, the P block, and the B block. In this case, the prediction candidate identification flag encoder 240 may generate and encode a prediction candidate selection flag differently in each case. In this case, in the prediction candidate selector 220 and the prediction candidate selection encoder 250, the motion vector prediction candidates selected differently according to the case of the P block and the B block in the block type of the SKIP mode are SKIP mode, respectively. Insert into and send. In the video decoding apparatus, when the block type of the current block is the SKIP mode, the P block, and the B block, the motion vector prediction candidates selected differently may be decoded and determined.

12 is a flowchart illustrating a motion vector encoding method according to an embodiment of the present invention.

The encoder 150 selects one or more motion vector prediction candidates (S1210), and determines one motion vector prediction candidate among the one or more motion vector prediction candidates as a prediction motion vector (S1220), and then, in step S1220, the current motion vector is determined. A difference vector is calculated by subtracting the predicted motion vector determined in S1230, and the calculated difference vector is encoded (S1240).

Here, the encoder 150 may generate and encode a prediction candidate identification flag that identifies which motion vector prediction candidate is determined as the prediction motion vector among the one or more motion vectors selected in step S1210. Also, since the encoder 150 selects one or more motion vector prediction candidates, determines a prediction motion vector, and generates and encodes a prediction candidate identification flag, the method described above with reference to FIGS. 2 to 11 will be described in detail. Is omitted. In addition, the order of steps shown in FIG. 12 is merely an example, and the present invention is not necessarily limited to this order, and the order may be selectively changed, and in some cases, some or all of the steps may be performed in parallel. It may be.

13 is a flowchart illustrating an image encoding method according to an embodiment of the present invention.

The image encoding apparatus 100 determines a current motion vector of the current block, generates a prediction block using the determined current motion vector (S1310), generates a residual block by subtracting the current block and the prediction block, and converts the residual block. Quantization and encoding (S1320). The image encoding apparatus 100 determines one of the motion vector prediction candidates as a predicted motion vector as a predicted motion vector (S1330) and encodes a difference vector obtained by subtracting the predicted motion vector determined in step S1330 from the current motion vector (S1340). In this case, the image encoding apparatus 100 may generate and encode a prediction candidate identification flag. The image encoding apparatus 100 generates and outputs a bitstream including the residual block encoded in step S1320 and the difference vector encoded in step S1340 (S1350). When the prediction candidate identification flag is encoded, the image encoding apparatus 100 may further include the encoded prediction candidate identification flag in the bitstream. In addition, the order of the steps shown in FIG. 13 is merely an example, and the present invention is not necessarily limited to this order, and the order may be selectively changed, and in some cases, some or all of the steps may be performed in parallel. It may be.

As described above, the image encoded in the bitstream by the video encoding apparatus 100 is connected to a wired or wireless communication network such as the Internet, a near field communication network, a wireless LAN network, a WiBro network, a mobile communication network, or the like in real time or in real time. The image decoding apparatus may be transmitted to an image decoding apparatus to be described later through a communication interface such as a universal serial bus (USB), and decoded by the image decoding apparatus to restore and reproduce the image.

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

The image decoding apparatus 1400 according to an embodiment of the present invention includes a decoder 1410, an inverse quantizer 1420, an inverse converter 1430, a predictor 1440, an adder 1450, and a memory 1460. Can be configured. The video decoding apparatus 1400 may be a personal computer (PC), a notebook computer, a personal digital assistant (PDA), a portable multimedia player (PMP), or a PlayStation Portable (PSP). ), 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 decoding an image, and executing a program. Means a variety of devices including a microprocessor for operation and control.

The decoder 1410 extracts and decodes the encoded residual block and the encoded difference vector from the bitstream, reconstructs the residual block and the difference vector, and converts one motion vector prediction candidate from one or more motion vector prediction candidates into a prediction motion vector. Determine and reconstruct the current motion vector of the current block by adding the predicted motion vector and the reconstructed difference vector.

Here, the decoder 1410 further extracts and decodes the encoded prediction candidate identification flag from the bitstream to further reconstruct the prediction candidate identification flag, and is identified by the reconstructed prediction candidate identification flag among one or more motion vector prediction candidates. The motion vector prediction candidate may be determined as the prediction motion vector.

In addition, the decoder 1410 decodes the encoded prediction candidate identification flag and the encoded partial prediction candidate identification flag from the bitstream to reconstruct the prediction candidate identification flag and the partial prediction candidate identification flag, and restores one or more motion vector prediction candidates. The motion vector prediction candidate identified by the reconstructed prediction candidate identification flag may be determined as the prediction motion vector among some motion vector prediction candidates identified by the predetermined prediction candidate identification flag.

Further, the decoder 1410 further extracts and decodes the encoded prediction candidate identification flag from the bitstream to further reconstruct the prediction candidate identification flag, but further extracts the absolute value and the y-component of the x-component of the reconstructed difference vector. If at least one of the absolute values is less than or equal to at least one of the preset x-component threshold and the preset y-component threshold, respectively, the x-component of the motion vector prediction candidate identified by the reconstructed prediction candidate identification flag. And one or more of the y-component may be determined as one or more of an x-component and a y-component of the predictive motion vector.

Inverse quantizer 1420 inverse quantizes the residual block reconstructed by decoder 1410, inverse transformer 1430 inverse transforms the inverse quantized residual block by inverse quantizer 1420, and predictor 1440 ) Generates a predictive block using the current motion vector reconstructed by the decoder 1410, and the adder 1450 predicts the residual block inversely transformed by the inverse transformer 1430 and the predictor 1440. Add the block to restore the current block. The memory 1460 stores the reconstructed current block output from the adder 1450 as a reference frame in units of frames so that the reference frame can be used by the predictor 1440.

Although not shown in FIG. 14, the image decoding apparatus 1400 according to an embodiment of the present invention described above is based on the H.264 / AVC standard, an intra predictor for intra prediction and a deblocking filtering of the reconstructed current block. And a deblocking filter for deblocking filtering.

15 is a block diagram schematically illustrating a motion vector decoding apparatus according to an embodiment of the present invention.

The motion vector decoding apparatus according to an embodiment of the present invention may be implemented as a decoder 1410 in the image decoding apparatus according to the embodiment of the present invention described above with reference to FIG. 14. Hereinafter, for convenience of description, the motion vector decoding apparatus according to an embodiment of the present invention is called a decoder 1410.

The decoder 1410 according to an embodiment of the present invention may include a differential vector decoder 1530 and a motion vector decompressor 1540. In addition, the decoder 1410 may further include a prediction candidate selection flag decoder 1510 and a prediction candidate identification flag decoder 1520 in some cases. In addition, the decoder 1410 may further include an image decoder that decodes the encoded residual block. However, since the image decoder is obvious to those skilled in the art, a detailed description thereof will be omitted.

The difference vector decoder 1530 reconstructs the difference vector by decoding the encoded difference vector. The motion vector reconstructor 1540 determines a motion vector prediction candidate of one or more motion vector prediction candidates as a prediction motion vector, and adds the determined prediction motion vector and the reconstructed difference vector to reconstruct the current motion vector of the current block. .

The prediction candidate selection flag decoder 1510 restores the encoded prediction candidate selection flag from the header of the bitstream, and the candidate identification flag decoder 1530 extracts and decodes one or more of the encoded prediction candidate identification flags from the bitstream. To restore. The motion vector decompressor 1540 may determine, as one or more motion vector prediction candidates, a motion vector prediction candidate identified by the prediction candidate selection flag reconstructed by the candidate selection flag decoder 1510. The motion vector prediction candidate identified by the prediction candidate identification flag reconstructed by 1520 may be determined as the prediction motion vector.

Here, when the candidate selection flag decoder 1510 is not present or the prediction candidate selection flag is not reconstructed by the candidate selection flag decoder 1510, the motion vector reconstructor 1530, like the encoder 150, of the current block may be used. One or more of a motion vector of the neighboring block, a motion vector of the Col block at the same position as the current block in the reference frame, a motion vector of the neighboring block of the Col block, and an instruction vector indicating a specific block of the reference frame in the current block. One or more motion vector prediction candidates may be selected. In this case, the specific block may be a block having a motion vector that is horizontal to the indication vector or smaller than a predetermined threshold. Since this is described above with reference to FIGS. 5 to 7, a detailed description thereof will be omitted.

In addition, when the selected one or more motion vector prediction candidates are all the same, it is assumed that the same motion vector prediction candidate is one motion vector prediction candidate, and the candidate identification flag decoder 1520 is determined as the prediction motion vector and reconstructed with the difference vector. In addition, the current motion vector may be restored.

In addition, the motion vector decompressor 1540 may set an x-component threshold value and a predetermined y- value in which the absolute value of the x-component and the y-component of the differential vector reconstructed by the differential vector decoder 1530 are preset. If it is larger than the component threshold value, the candidate identification flag does not need to be decoded, and the motion vector prediction candidate having the minimum current motion vector reconstructed using each of the one or more motion vector prediction candidates may be determined as the prediction motion vector.

8 and 9, for example, when the difference vector is larger than the threshold, the motion vector reconstructor 1540 selects a motion vector prediction candidate from among the motion vector prediction candidates selected without the prediction candidate identification flag. If it is not possible to determine whether the current motion vector is reconstructed with the x-component of the first motion vector prediction candidate, the current motion vector becomes '7' (10-3 = 7). However, since the x-component of the second motion vector prediction candidate is '7', it may be identified that it is not encoded using the x-component of the first motion vector prediction candidate. If the current motion vector is reconstructed using the x-component of the second motion vector prediction candidate, the current motion vector is '4' (7-3 = 4), so if the current motion vector is '4', the second motion vector It can be identified that the case predicted by the third motion vector prediction candidate is more optimal than that predicted by the x-component of the prediction candidate. Thus, it can be seen that it is not coded using the x-component of the second motion vector prediction candidate. Through this method, since the x-component of the third motion vector prediction candidate is optimal, the image encoding apparatus 100 determines the x-component of the prediction motion vector as the x-component of the third motion vector prediction candidate. Can be identified.

In addition, when the decoder 1410 does not need to decode the prediction candidate identification flag because the absolute value of each component of the difference vector is larger than the corresponding component threshold, the motion vector prediction candidates are separately decoded for each component to decode the prediction motion vector. You can decide. For example, as described above in the description of the encoder 150, when the absolute values of two components of the difference vector are both greater than the threshold of each component, the optimal prediction motion vector for the x-component is 'A'. The optimal predictive motion vector of the y-component may be determined as 'B', and the predictive motion vector may be determined as a motion vector prediction candidate for each component.

In addition, when the absolute value of each component of the difference vector is smaller than the corresponding component threshold or only one of the two components, the decoder 1410 must decode the prediction candidate identification flag. The motion vector reconstructor 1540 may determine a motion vector prediction candidate identified by the prediction candidate identification flag reconstructed by the prediction candidate identification flag decoder 1520 as the motion vector prediction candidate. Further, when both the x-component and the y-component of the difference vector are smaller than the threshold value, the motion vector prediction candidate identified by decoding the prediction candidate identification flag may be determined as the prediction motion vector, respectively, for the x-component and the y-component. One prediction candidate identification flag may be decoded to determine a motion vector prediction candidate that satisfies both components simultaneously as the prediction motion vector. When only one of the absolute values of two components of the difference vector is larger than the threshold of the component, the prediction candidate identification flag is decoded only for the component smaller than the threshold of the component. The motion vector decompressor 1540 is configured such that at least one of an absolute value of the x-component and an absolute value of the y-component of the difference vector reconstructed by the differential vector decoder 1530 is a preset x-component threshold and a preset value. If less than or equal to one or more of the y-component thresholds, one of the x-component and the y-component of the motion vector prediction candidate identified by the prediction candidate identification flag reconstructed by the prediction candidate identification flag decoder 1520. The anomaly may be determined as one or more of an x-component and a y-component of the predicted motion vector. In this case, when the prediction candidate selection flag decoder 1510 restores the prediction candidate selection flag, the motion vector decompressor 1540 may reconstruct the prediction candidate selection flag from among the motion vector prediction candidates identified by the prediction candidate selection flag. It goes without saying that one or more of the x-component and y-component of the identified motion vector prediction candidate can be determined as one or more of the x-component and y-component of the predictive motion vector.

For example, the prediction candidate identification flag encoder 240 described above with reference to FIG. 2 is a prediction that identifies which motion vector prediction candidate is used as one component larger than the threshold when only one component is smaller than the threshold. A candidate identification flag is generated, and if two components are not greater than the threshold, one prediction candidate identification flag is generated and coded by the two components to identify which motion vector prediction candidate was used, or two components Two prediction candidate identification flags that identify whether a motion vector prediction candidate is used may be generated and encoded.

Accordingly, when the prediction candidate identification flag decoder 1520 extracts and decodes two encoded prediction candidate identification flags from the bitstream, the motion vector reconstructor 1540 may determine the motion vector prediction candidate identified by the two prediction candidate identification flags. The x-component and y-component of may be determined as the x-component and y-component of the predicted motion vector. If the candidate identification flag decoder 1520 extracts and decodes one encoded prediction candidate identification flag, the motion vector reconstructor 1540 obtains the x-direction of the motion vector prediction candidate identified by one prediction candidate identification flag, Component or y-component of the predicted motion vector or a x-component or y-component of the motion vector prediction candidate identified by one predictive candidate identification flag to the x-component of the predicted motion vector Or a y-component.

In this case, the motion vector reconstructor 1540 determines which component of the x-component and the y-component of the motion vector prediction candidate identified by one prediction candidate identification flag as the corresponding component of the prediction motion vector. ) May be determined as a component that is promised (or preset) with the prediction candidate identification flag encoder 240. In addition, when the motion vector reconstructor 1540 determines any component of the x-component and the y-component of the motion vector prediction candidate identified by one prediction candidate identification flag as the corresponding component of the prediction motion vector, the motion vector reconstructor 1540 The other component may determine, as the corresponding component of the prediction motion vector, a corresponding component of the motion vector prediction candidate having the minimum current motion vector reconstructed using each of the corresponding components of the one or more motion vector prediction candidates.

In addition, when the decoder 1410 uses N motion vector prediction candidates selected by the prediction candidate selection flag decoder 1510 as N, the number of prediction candidates selected as shown in Equation 7 in the description of the encoder 150 is N. The lengths of the prediction candidate identification flags to be decoded vary according to the magnitude of the difference vector as shown in FIGS. 8 and 10. The prediction candidate identification flag decoder 1520 generates and decodes the prediction candidate identification flag differently according to the magnitude of the absolute value of each component of the difference vector and the threshold of each component. That is, the prediction candidate identification flag decoder 1520 sorts the N-1 thresholds in descending order so that when the absolute value of each component of the difference vector is less than or equal to the smallest threshold value of the corresponding component, the N selected motion vectors Decoding a prediction candidate identification flag identifying which motion vector prediction candidate among the prediction candidates is determined as the prediction motion vector, and if the absolute value of each component of the difference vector is less than or equal to the second smallest threshold of that component, Assume that there is no current motion vector between the two motion vector prediction candidates having the smallest threshold of the component, and identify which motion vector prediction candidate is determined as the prediction motion vector among the N-1 selected motion vector prediction candidates. The prediction candidate identification flag is decoded. If the absolute value of each component of the difference vector is less than or equal to the third smallest threshold of the component, the assumption is that there is no current motion vector between the two motion vector prediction candidates with the smallest threshold of the component. Assuming that there is no current motion vector between the two motion vector prediction candidates having the second smallest threshold of the component, a motion vector prediction candidate is determined as the prediction motion vector among the N-2 selected motion vector prediction candidates. A prediction candidate identification flag for identifying whether the answer is correct.

The length of the candidate prediction candidate identification flag to be decoded is the highest probability motion vector prediction candidate when the value calculated by Equation 9 is an integer as described above in the encoder as shown in FIG. Is determined. In case of '0', the motion vector prediction candidate is determined by reading the maximum length of the prediction candidate identification flag as shown in Equation (8). If the value calculated by Equation 9 is not an integer, the motion vector prediction candidate MVp_cand [i] is determined by reading the maximum length of the prediction candidate identification flag as shown in Equation 8.

8 and 10, for example, the number of motion vector prediction candidates is four and each threshold is '1', '3', or '5'. 10 is a diagram of dividing sections according to three threshold values. The interval denoted by ① indicates the x-components of the motion vector prediction candidates, and the interval denoted by ② denotes the current motion vector when the absolute value of the x-component of the difference vector is less than or equal to the smallest threshold of the x-component. where the x-component is located, and the section marked with ③ is the x-component of the current motion vector when the absolute value of the x-component of the difference vector is less than or equal to the second smallest threshold of the x-component. The interval indicated by ④ indicates the interval in which the x-component of the current motion vector is located when the absolute value of the x-component of the difference vector is less than or equal to the largest threshold of the x-component. , The interval ⑤ denotes the x-component of the current motion vector when the absolute value of the x-component of the difference vector is greater than the largest threshold of the x-component. A shows a section in which location.

The prediction candidate identification flag decoder 1520 decodes a prediction candidate identification flag that identifies which motion vector prediction candidate is determined as the prediction motion vector among the four motion vector prediction candidates in the section denoted by ②. Decoding a prediction candidate identification flag that identifies which motion vector prediction candidate is determined as the prediction motion vector among the three motion vector prediction candidates, and predicts which motion vector prediction candidate is among the two motion vector prediction candidates in the section denoted by ④. The prediction candidate identification flag for identifying whether or not it is determined as a vector is decoded, and the prediction candidate identification flag is not decoded in the section denoted by 5.

When the difference vector is '+4', since it is larger than the second largest threshold and smaller than the largest threshold, it can be seen that it belongs to the section ④ of FIG. 10 as described above in the description of the encoder 150. have. The prediction candidate identification flag indicating which motion vector prediction candidate is used by the encoder 150 among '4' and '14' may be decoded. In this case, since the number of motion vector prediction candidates to be decoded by Equation 8 is two, only one bit needs to be decoded. When the difference vector is '-3', it belongs to the section 3 of FIG. 10, and thus, a prediction candidate indicating which prediction candidate is used by the encoder 150 among the motion vector prediction candidates '-4', '2', and '14'. The identification flag may be decrypted. In this case, since the number of prediction candidates to be decoded is 3, the value calculated by Equation 9 becomes an integer. Thus, when only one bit is first read as shown in FIG. 11, MVp_cand [0] is a motion vector. When used as a prediction candidate, and in the case of '0', it may be decoded by the length of the prediction candidate identification flag to be decoded by Equation (8).

In addition, when the decoder 1410 further includes the prediction candidate identification flag decoder 1520, the motion vector decompressor 1540 reads only one bit of the prediction candidate identification flag, so that one bit read is, for example. It is determined whether or not it is '1' to identify whether the motion vector prediction candidate having the highest probability is determined as the predictive motion vector, and only when one or more bits are selected if the motion vector prediction candidate having the highest probability is not determined as the prediction motion vector. Further reading may identify which motion vector prediction candidate was determined as the predicted motion vector.

In this case, the motion vector reconstructor 1540 may calculate the probability of determining the current motion vector for each of the one or more motion vector prediction candidates by encoding one or more regions of the current frame using one or more motion vector prediction candidates. A motion vector prediction candidate that is designated or randomly designated according to a designated probability may be selected as a motion vector prediction candidate having the highest probability.

16 is a flowchart illustrating a motion vector decoding method according to an embodiment of the present invention.

The decoder 1410 reads and decodes the encoded prediction candidate identification flag from the bitstream to restore the prediction candidate identification flag (S1610), reads and decodes the encoded difference vector from the bitstream to restore the differential vector (S1620), The encoded prediction candidate identification flag is read and decoded from the bitstream to restore the prediction candidate identification flag (S1630), and among the one or more motion vector prediction candidates, the motion vector prediction candidate identified by the prediction candidate identification flag reconstructed in step S1630 is predicted. A motion vector is determined, and the determined prediction motion vector and the difference vector reconstructed in step S1620 are added to reconstruct the current motion vector of the current block (S1640). Here, the decoder 1410 may not necessarily perform steps S1610 and S1630. In this case, when the decoder 1410 determines the prediction motion vector in step S1640, as described above with reference to FIG. If there is no flag, the prediction motion vector may be determined according to a method of determining the prediction motion vector. In addition, the order of the steps shown in FIG. 16 is merely an example, and the present invention is not necessarily limited to this order, and the order may be selectively changed.

17 is a flowchart illustrating an image decoding method according to an embodiment of the present invention.

The image decoding apparatus 1400 restores the prediction candidate identification flag by extracting and decoding the encoded prediction candidate identification flag from the bitstream (S1710), and restores the difference vector by extracting and decoding the encoded difference vector from the bitstream ( S1720, and reconstructing the prediction candidate identification flag by extracting and decoding the encoded prediction candidate identification flag from the bitstream (S1730), and among the one or more motion vector prediction candidates, the motion vector prediction candidate by the prediction candidate identification flag reconstructed in step S1730. Is determined as the predicted motion vector, and the predicted motion vector and the difference vector reconstructed in step S1720 are added to reconstruct the current motion vector (S1740).

The image decoding apparatus 1400 estimates and compensates for the current block by generating and predicting the motion of the current block using the current motion vector reconstructed in step S1740 (S1750), and extracts the encoded residual block from the bitstream. The encoded residual block is inversely quantized, inversely transformed, and decoded to restore the residual block, and the reconstructed residual block is added to the prediction block to restore the current block (S1760).

Here, the image decoding apparatus 1400 may not necessarily perform steps S1710 and S1730. In this case, when the image decoding apparatus 1400 determines the prediction motion vector in step S1740, as described above with reference to FIG. 15, the prediction may be performed. If there is no candidate identification flag, the prediction motion vector may be determined according to a method of determining the prediction motion vector. In addition, the steps shown and described in FIG. 17 are not necessarily limited to the order shown and described, and the order may be selectively changed.

For example, the image decoding apparatus 1400 does not perform each step in the order of FIG. 17, but extracts and decodes the encoded residual block from the bitstream to restore the residual block, and decodes the encoded difference vector from the bitstream. Extracting and decoding to reconstruct the difference vector, selecting one motion vector prediction candidate among the one or more motion vector prediction candidates as the prediction motion vector, and reconstructing the current motion vector of the current block by using the selected prediction motion vector and the reconstructed difference vector. Inversely quantizes the reconstructed residual block, inversely transforms the inverse quantized residual block, generates a prediction block using the reconstructed current motion vector, and adds the inverse transformed residual block and the generated prediction block to reconstruct the current block. Each step can be performed in the order that

While the present invention has been described in connection with what is presently considered to be the most practical and preferred embodiment, it is to be understood that the invention is not limited to the disclosed embodiments. In other words, within the scope of the present invention, all of the components may be selectively operated in combination with one or more. In addition, although all of the components may be implemented as one independent hardware, some or all of the components may be selectively combined to perform a part or all of the functions in one or a plurality of hardware. As shown in FIG. Codes and code segments constituting the computer program may be easily inferred by those skilled in the art. Such a computer program may be stored in a computer readable storage medium and read and executed by a computer, thereby implementing embodiments of the present invention. As the storage medium of the computer program, a magnetic recording medium, an optical recording medium, a carrier wave medium, or the like may be included.

Furthermore, the terms "comprises", "comprising", or "having" described above mean that a component can be implanted unless otherwise specifically stated, But should be construed as including other elements. All terms, including technical and scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs, unless otherwise defined. Terms used generally, such as terms defined in a dictionary, should be interpreted to coincide with the contextual meaning of the related art, and shall not be interpreted in an ideal or excessively formal sense unless explicitly defined in the present invention.

The foregoing description is merely illustrative of the technical idea of the present invention, and various changes and modifications may be made by those skilled in the art without departing from the essential characteristics of the present invention. Therefore, the embodiments disclosed in the present invention are intended to illustrate rather than limit the scope of 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.

As described above, the present invention performs inter prediction using a pixel that has been encoded, decoded, and reconstructed before a block to be currently encoded, and efficiently encodes or decodes a motion vector used for inter prediction, and then uses a still image or Applied to the field of image processing that encodes or decodes the data of a video, it is possible to improve the compression efficiency by reducing the amount of bits for encoding the information on the predicted motion vector while encoding the motion vector using the predicted motion vector more accurately. It is a very useful invention that produces an effect.

Claims (12)

An apparatus for encoding a current motion vector which is a motion vector of a current block,
A prediction candidate selector for selecting a motion vector prediction candidate for predicting the current motion vector;
A predicted motion vector determiner that determines a predicted motion vector from the motion vector prediction candidate; And
And a decoder for calculating a difference vector by subtracting the predicted motion vector from the current motion vector, and encoding the differential vector and prediction candidate identification information for identifying the determined predicted motion vector,
The prediction candidate selector,
A module for selecting a first motion vector prediction candidate from a neighboring block of the current block, and a second motion vector from a Col block or a neighboring block of the Col block at a position corresponding to the position of the current block in a reference frame; And a module for selecting a prediction candidate. The method of claim 1,
The module for selecting the first motion vector prediction candidate,
And selecting the first motion vector prediction candidate from neighboring blocks at a predetermined position among all neighboring blocks of the current block. The method of claim 1,
The prediction candidate selector,
And a distance between two motion vector prediction candidates does not satisfy a predetermined criterion, and selects only one of them as a motion vector prediction candidate. In the method of encoding the current motion vector which is a motion vector of the current block,
Selecting a motion vector prediction candidate for predicting the current motion vector;
Determining a prediction motion vector from the motion vector prediction candidate; And
Subtracting the prediction motion vector from the current motion vector to calculate a difference vector, and encoding prediction candidate identification information and the difference vector for identifying the determined prediction motion vector,
The motion vector prediction candidate selection step,
Selecting a first motion vector prediction candidate in a neighboring block of the current block; and a Collocated block or a second motion vector in a neighboring block of the Col block at a position corresponding to the position of the current block in a reference frame. Selecting a prediction candidate. 5. The method of claim 4,
Selecting the first motion vector prediction candidate,
And selecting the first motion vector prediction candidate from neighboring blocks at a predetermined position among all neighboring blocks of the current block. 5. The method of claim 4,
Selecting the motion vector prediction candidate,
If the distance between two motion vector prediction candidates does not satisfy a predetermined criterion, only one of the motion vector prediction candidates is selected as the motion vector prediction candidate. In the apparatus for decoding the current motion vector of the current block,
A decoder for decoding the differential vector and the prediction candidate identification information from the bitstream; And
Selecting a motion vector prediction candidate for predicting the current motion vector, determining a motion vector prediction code identified by the prediction candidate identification information among the motion vector prediction candidates as a prediction motion vector, and comparing the difference with the prediction motion vector A motion vector decompressor for adding the vector to restore the current motion vector of the current block;
The motion vector decompressor,
A module for selecting a first motion vector prediction candidate from a neighboring block of the current block, and a second motion vector from a Col block or a neighboring block of the Col block at a position corresponding to the position of the current block in a reference frame; And a module for selecting a prediction candidate. The method of claim 7, wherein
The motion vector decompressor,
And selecting the first motion vector prediction candidate from neighboring blocks at a predetermined position among all neighboring blocks of the current block. The method of claim 7, wherein
The motion vector decompressor,
And selecting only one of the motion vector prediction candidates when the distance between the two motion vector prediction candidates does not satisfy a predetermined criterion. In the method of decoding the current motion vector of the current block,
Recovering the difference vector from the bitstream;
Selecting a motion vector prediction candidate for predicting the current motion vector;
Restoring prediction candidate identification information from the bitstream;
Determining a motion vector prediction candidate identified by the prediction candidate identification information among the motion vector prediction candidates as a prediction motion vector; And
Reconstructing the current motion vector of the current block by adding the prediction motion vector and the difference vector;
The motion vector prediction candidate selection step,
Selecting a first motion vector prediction candidate from a neighboring block of the current block; and a second motion vector from a Col block or a neighboring block of the Col block at a position corresponding to the position of the current block in a reference frame. Selecting a prediction candidate. 11. The method of claim 10,
The first motion vector prediction candidate selection step
And selecting the first motion vector prediction candidate from neighboring blocks at a predetermined position among all neighboring blocks of the current block. 11. The method of claim 10,
The motion vector prediction candidate selection step,
If the distance between two motion vector prediction candidates does not satisfy a predetermined criterion, only one of the motion vector prediction candidates is selected as the motion vector prediction candidate.

Images