KR102503694B1 - Method for inter prediction and apparatus thereof - Google Patents

Abstract

The inter prediction method according to the present invention includes determining at least one candidate search position for a current block, selecting an initial search position from among at least one candidate search position, and selecting a search position within a set search range based on the initial search position. Deriving a motion vector of the current block by performing motion prediction, and generating a prediction block by performing prediction on the current block based on the derived motion vector.

Description

Inter prediction method and its device {METHOD FOR INTER PREDICTION AND APPARATUS THEREOF}

The present invention relates to image processing, and more particularly, to a motion estimation method and apparatus.

Recently, as broadcasting services with high definition (HD) resolution are expanded not only domestically but also globally, many users are getting used to high-resolution and high-definition images, and many organizations are accelerating the development of next-generation video equipment. In addition, as interest in UHD (Ultra High Definition) having a resolution four times higher than that of HDTV increases along with HDTV, a compression technology for a higher resolution and higher quality image is required.

For image compression, inter prediction technology predicts pixel values included in the current picture from temporally previous and/or subsequent pictures, and predicts pixel values included in the current picture using pixel information in the current picture. An intra prediction technique, an entropy coding technique in which short codes are assigned to symbols with a high frequency of occurrence and long codes are assigned to symbols with a low frequency of occurrence may be used.

A technical problem of the present invention is to provide an image encoding method and apparatus capable of improving image encoding performance.

Another technical problem of the present invention is to provide an inter prediction method and apparatus capable of improving video encoding performance.

A technical problem of the present invention is to provide a motion prediction method and apparatus capable of improving video encoding performance.

One embodiment of the present invention is a motion prediction method. The method includes determining at least one candidate search position for a current block, selecting an initial search position from among the at least one candidate search position, and predicting motion within a set search range based on the initial search position. and deriving a motion vector of the current block by performing At this time, in the step of selecting the initial search position, the initial search position may be selected based on the encoding cost of the at least one candidate search position.

The current block may be one of a plurality of lower blocks generated by dividing an upper block for which motion prediction has already been completed, and the at least one candidate search position is a motion vector of the upper block based on a zero point position of the current block. may include the location indicated by

The current block may be one of a plurality of sub-blocks generated by dividing an upper block for which motion prediction has already been completed, and the at least one candidate search position is based on a zero point position of the current block, and the plurality of sub-blocks Among them, it may be a position indicated by a motion vector of a block for which motion prediction has already been completed.

The at least one candidate search position may include a position indicated by a motion vector of a co-located block in a reference picture to be used for inter prediction of the current block, based on the zero point position of the current block, wherein the co-located block is the It may be a block existing in the same spatial position as the current block in the reference picture.

The at least one candidate retrieval position may further include a position indicated by a motion vector of a block located adjacent to the co-located block in the reference picture based on the position of the zero point of the current block.

The at least one candidate search position may include a position indicated by a combined motion vector derived based on a plurality of motion vectors based on the position of the zero point of the current block. Here, each of the plurality of motion vectors may be a motion vector of a block for which motion prediction has been completed.

The current block may be one of a plurality of lower blocks generated by dividing an upper block for which motion prediction has already been completed, and the plurality of motion vectors include an origin vector indicating the position of the zero point, a motion vector of the upper block, and the plurality of motion vectors. Among a plurality of sub-blocks, at least one of a motion vector of a block for which motion prediction has been completed, a predicted motion vector of the current block, and a motion vector of a block adjacent to the current block may be included.

The combined motion vector may be derived by averaging the plurality of motion vectors.

The combined motion vector may be derived by a weighted sum of the plurality of motion vectors.

The X component value of the combined motion vector may be determined as the maximum value among the X component values of the plurality of motion vectors, and the Y component value of the combined motion vector may be determined as the maximum value among the Y component values of the plurality of motion vectors. there is.

The X component value of the combined motion vector may be determined as the minimum value among the X component values of the plurality of motion vectors, and the Y component value of the combined motion vector may be determined as the minimum value among the Y component values of the plurality of motion vectors. there is.

The selecting of the initial search positions may include determining a predetermined number of final candidate search positions from among the at least one candidate search position based on a degree of correlation between motion vectors representing the at least one candidate search position; and The method may further include selecting the initial search position from among the predetermined number of final candidate search positions.

The at least one candidate search position may include a position indicated by a predicted motion vector of the current block based on the zero point position of the current block, and in the step of determining the final candidate search position, the at least one candidate search position may be included. The final candidate search position may be determined based on motion vectors other than the predicted motion vector among motion vectors indicating a position and a difference value between the predicted motion vectors.

The current block may be one of a plurality of sub-blocks generated by dividing an upper block for which motion prediction has already been completed, and the at least one candidate search position is based on the position of the zero point of the current block for the upper block. It may include a position indicated by an upper motion vector generated by motion prediction, and in the step of determining the final candidate search position, the remaining motion vectors excluding the upper motion vector among the motion vectors indicating the at least one candidate search position , and the final candidate search position may be determined based on a difference value between the upper motion vectors.

The current block may be one of a plurality of lower blocks generated by dividing an upper block for which motion prediction has already been completed, and the at least one candidate search position is for a block for which motion prediction has already been completed among the plurality of lower blocks. It may include a position indicated by a lower motion vector generated by motion prediction, and in the step of determining the final candidate search position, the remaining motion vectors excluding the lower motion vector among the motion vectors indicating the at least one candidate search position , and the final candidate search position may be determined based on a difference value between the lower motion vectors.

In the step of determining the final candidate search position, the final candidate search position may be determined based on a variance value of each motion vector indicating the at least one candidate search position.

Another embodiment of the present invention is an inter prediction method. The method may include determining at least one candidate search position for a current block, selecting an initial search position from among the at least one candidate search position, and performing motion estimation within a set search range based on the initial search position. and deriving a motion vector of the current block by performing, and generating a prediction block by performing prediction on the current block based on the derived motion vector. At this time, in the step of selecting the initial search position, the initial search position may be selected based on the encoding cost of the at least one candidate search position.

Another embodiment of the present invention is an inter prediction device. The apparatus determines at least one candidate search position for a current block, selects an initial search position from among the at least one candidate search position, and performs motion prediction within a set search range based on the initial search position. It may include a motion estimation unit that derives a motion vector of the current block, and a motion compensation unit that generates a predicted block by performing prediction on the current block based on the derived motion vector. Here, the motion estimation unit may select the initial search position based on the coding cost of the at least one candidate search position.

Another embodiment of the present invention is a video encoding method. The method may include determining at least one candidate search position for a current block, selecting an initial search position from among the at least one candidate search position, and performing motion estimation within a set search range based on the initial search position. Deriving a motion vector of the current block by performing, generating a prediction block by performing prediction on the current block based on the derived motion vector, and generating a residual block based on the current block and the prediction block. and encoding the residual block. At this time, in the step of selecting the initial search position, the initial search position may be selected based on the encoding cost of the at least one candidate search position.

According to the video encoding method according to the present invention, video encoding performance can be improved.

According to the inter prediction method according to the present invention, video encoding performance can be improved.

According to the motion prediction method according to the present invention, video encoding performance can be improved.

1 is a block diagram showing a configuration according to an embodiment of an image encoding apparatus to which the present invention is applied.
2 is a block diagram showing a configuration of an image decoding apparatus according to an embodiment.
3 is a flowchart schematically illustrating an embodiment of an inter prediction method.
4 is a flowchart schematically illustrating an embodiment of a motion estimation process to which the present invention is applied.
5 is a diagram schematically illustrating an embodiment of a method for determining an initial search location.
6 is a diagram schematically illustrating an embodiment of a method for determining a candidate search position according to the present invention.
7 is a diagram schematically illustrating another embodiment of a method for determining a candidate search position according to the present invention.
8 is a diagram schematically illustrating another embodiment of a method for determining a candidate search position according to the present invention.

EMBODIMENT OF THE INVENTION Hereinafter, embodiment of this invention is described concretely with reference to drawings. In describing the embodiments of the present specification, if it is determined that a detailed description of a related known configuration or function may obscure the gist of the present specification, the detailed description thereof will be omitted.

It is understood that when a component is referred to as “connected” or “connected” to another component, it may be directly connected or connected to the other component, but other components may exist in the middle. It should be. In addition, the description of "including" a specific configuration in the present invention does not exclude configurations other than the corresponding configuration, and means that additional configurations may be included in the practice of the present invention or the scope of the technical spirit of the present invention.

Terms such as first and second may be used to describe various components, but the components should not be limited by the terms. These terms are only used for the purpose of distinguishing one component from another. For example, a first element may be termed a second element, and similarly, a second element may be termed a first element, without departing from the scope of the present invention.

In addition, components appearing in the embodiments of the present invention are shown independently to represent different characteristic functions, and do not mean that each component is made of separate hardware or a single software component unit. That is, each component is listed and included as each component for convenience of explanation, and at least two components of each component can be combined to form a single component, or one component can be divided into a plurality of components to perform a function, and each of these components can be divided into a plurality of components. Integrated embodiments and separated embodiments of components are also included in the scope of the present invention as long as they do not depart from the essence of the present invention.

In addition, some of the components may be optional components for improving performance rather than essential components that perform essential functions in the present invention. The present invention can be implemented by including only components essential to implement the essence of the present invention, excluding components used for performance improvement, and a structure including only essential components excluding optional components used for performance improvement. Also included in the scope of the present invention.

1 is a block diagram showing a configuration according to an embodiment of an image encoding apparatus to which the present invention is applied.

Referring to FIG. 1 , the video encoding apparatus 100 includes a motion estimator 111, a motion compensation unit 112, an intra prediction unit 120, a switch 115, a subtractor 125, and a transform unit. 130, quantization unit 140, entropy encoding unit 150, inverse quantization unit 160, inverse transform unit 170, adder 175, filter unit 180, and reference picture buffer 190. .

The image encoding apparatus 100 may encode an input image in an intra mode or an inter mode and output a bitstream. Intra prediction means prediction within a picture, and inter prediction means prediction between pictures. In the case of the intra mode, the switch 115 may be switched to the intra mode, and in the case of the inter mode, the switch 115 may be switched to the inter mode. After generating a prediction block for an input block of an input image, the image encoding apparatus 100 may encode a residual between the input block and the prediction block.

In the case of the intra mode, the intra prediction unit 120 may generate a prediction block by performing spatial prediction using pixel values of previously encoded blocks adjacent to the current block.

In the case of the inter mode, the motion estimator 111 may obtain a motion vector by finding a region that best matches the input block in the reference picture stored in the reference picture buffer 190 during the motion estimation process. there is. The motion compensator 112 may generate a prediction block by performing motion compensation using a motion vector.

The subtractor 125 may generate a residual block by a difference between the input block and the generated prediction block. The transform unit 130 may output a transform coefficient by performing transform on the residual block. The quantization unit 140 may quantize the input transform coefficient according to a quantization parameter and output a quantized coefficient.

The entropy encoding unit 150 performs entropy encoding based on the values (eg, quantized coefficients) calculated by the quantization unit 140 and/or the encoding parameter values calculated in the encoding process to generate a bitstream ( bit stream) can be output.

When entropy encoding is applied, a small number of bits are allocated to a symbol with a high probability of occurrence and a large number of bits are allocated to a symbol with a low probability of occurrence to represent the symbol, thereby reducing the number of bits for symbols to be coded. The size of the column may be reduced. Therefore, compression performance of image encoding can be improved through entropy encoding. The entropy encoding unit 150 may use an encoding method such as exponential golomb, context-adaptive variable length coding (CAVLC), or context-adaptive binary arithmetic coding (CABAC) for entropy encoding.

Since the video encoding apparatus according to the embodiment of FIG. 1 performs inter-prediction encoding, that is, inter-frame prediction encoding, a currently encoded video needs to be decoded and stored to be used as a reference video. Therefore, the quantized coefficient is inversely quantized in the inverse quantization unit 160 and inversely transformed in the inverse transformation unit 170. The inverse quantized and inverse transformed coefficients are added to the prediction block through an adder 175, and a reconstructed block is generated.

The reconstructed block passes through a filter unit 180, and the filter unit 180 applies at least one of a deblocking filter, a sample adaptive offset (SAO), and an adaptive loop filter (ALF) to the reconstructed block or the reconstructed picture. can do. The filter unit 180 may also be called an adaptive in-loop filter. The deblocking filter may remove block distortion and/or blocking artifacts generated at the boundary between blocks. SAO may add an appropriate offset value to a pixel value to compensate for a coding error. ALF may perform filtering based on a value obtained by comparing a reconstructed image with an original image, and may be performed only when high efficiency is applied. A reconstructed block that has passed through the filter unit 180 may be stored in the reference picture buffer 190 .

2 is a block diagram showing a configuration of an image decoding apparatus according to an embodiment.

Referring to FIG. 2 , the image decoding apparatus 200 includes an entropy decoding unit 210, an inverse quantization unit 220, an inverse transform unit 230, an intra prediction unit 240, a motion compensation unit 250, an adder ( 255), a filter unit 260, and a reference picture buffer 270.

The image decoding apparatus 200 may receive the bitstream output from the encoder, perform decoding in an intra mode or inter mode, and output a reconstructed image, that is, a reconstructed image. In the intra mode, the switch may be converted to intra, and in the case of the inter mode, the switch may be converted to inter. The image decoding apparatus 200 may generate a reconstructed block, that is, a reconstructed block, by obtaining a residual block from the input bitstream, generating a prediction block, and then adding the residual block and the prediction block.

The entropy decoding unit 210 may generate symbols including symbols in the form of quantized coefficients by entropy decoding the input bitstream according to a probability distribution. The entropy decoding method is similar to the entropy encoding method described above.

When the entropy decoding method is applied, a small number of bits are allocated to a symbol with a high probability of occurrence and a large number of bits are allocated to a symbol with a low probability of occurrence to express the symbol, thereby increasing the size of the bit stream for each symbol. can be reduced Therefore, compression performance of image decoding may be improved through the entropy decoding method.

The quantized coefficient is inversely quantized in the inverse quantization unit 220 and inversely transformed in the inverse transform unit 230, and as a result of inverse quantization/inverse transformation of the quantized coefficient, a residual block may be generated.

In the intra mode, the intra prediction unit 240 may generate a prediction block by performing spatial prediction using pixel values of previously decoded blocks adjacent to the current block. In the case of the inter mode, the motion compensation unit 250 may generate a prediction block by performing motion compensation using a motion vector and a reference image stored in the reference picture buffer 270 .

The residual block and the prediction block are added through the adder 255, and the added block may pass through the filter unit 260. The filter unit 260 may apply at least one of a deblocking filter, SAO, and ALF to a reconstructed block or a reconstructed picture. The filter unit 260 may output a reconstructed image, that is, a restored image. The reconstructed image may be stored in the reference picture buffer 270 and used for inter prediction.

Hereinafter, a block means a unit of video encoding and decoding. When encoding and decoding an image, the encoding or decoding unit refers to the divided unit when encoding or decoding an image by dividing it, such as a coding unit (CU), a prediction unit (PU), and a transform unit (TU). : Transform Unit), a transform block, and the like. One block may be further divided into smaller sub-blocks.

3 is a flowchart schematically illustrating an embodiment of an inter prediction method.

Referring to FIG. 3, an encoder and a decoder may derive motion information for a current block (S310).

In the inter mode, the encoder and decoder may derive motion information of the current block and then perform inter prediction and/or motion compensation based on the derived motion information. The encoder may derive motion information on the current block by performing motion estimation on the current block. At this time, the encoder may transmit information related to the motion information to the decoder. The decoder may derive motion information of the current block based on information transmitted from the encoder. Specific embodiments of a motion prediction method for the current block will be described later.

In this case, the encoder and the decoder can improve encoding/decoding efficiency by using motion information of a 'Col block' corresponding to the current block in a reconstructed neighboring block and/or a previously reconstructed 'Col picture'. can Here, the reconstructed neighboring block is a block in the current picture that has already been encoded and/or decoded and reconstructed, and may include a block adjacent to the current block and/or a block located at an outer corner of the current block. In addition, the encoder and decoder may determine a predetermined relative position based on a block located at the same spatial position as the current block in the Col picture, and the determined predetermined relative position (existing at the same spatial position as the current block) The Col block may be derived based on the location inside and/or outside the block to be selected. Here, for example, the Col picture may correspond to one picture among reference pictures included in the reference picture list.

Meanwhile, a motion information encoding method and/or a motion information derivation method may vary according to a prediction mode of a current block. Prediction modes applied for inter prediction may include advanced motion vector prediction (AMVP), merge, and the like.

For example, when Advanced Motion Vector Prediction (AMVP) is applied, the encoder and the decoder generate a predicted motion vector candidate list using motion vectors of reconstructed neighboring blocks and/or motion vectors of Col blocks. can create That is, a motion vector of a reconstructed neighboring block and/or a motion vector of a Col block may be used as a predicted motion vector candidate. The encoder may transmit a predicted motion vector index indicating an optimal predicted motion vector selected from among the predicted motion vector candidates included in the list to the decoder. In this case, the decoder may select a motion vector predictor of the current block from among motion vector predictor candidates included in the predictor motion vector candidate list by using the predictor motion vector index.

Hereinafter, in this specification, for convenience of description, the predicted motion vector candidate may be referred to as a predicted motion vector (PMV), and the predicted motion vector may also be referred to as a motion vector predictor (MVP). Such distinction will be easily made by those skilled in the art.

The encoder may obtain a motion vector difference (MVD) corresponding to a difference between the motion vector of the current block and the predicted motion vector of the current block, and may encode and transmit the motion vector difference (MVD) to the decoder. At this time, the decoder may decode the received motion vector difference and derive a motion vector of the current block through the sum of the decoded motion vector difference and the predicted motion vector.

Meanwhile, the encoder and the decoder use the median value of the motion vectors of the reconstructed neighboring blocks as the predicted motion vector instead of using the motion vector of the reconstructed neighboring block and/or the motion vector of the Col block as the predicted motion vector. may be In this case, the encoder may encode and decode the difference between the motion vector value of the current block and the median value, and transmit the result to the decoder. At this time, the decoder may decode the received difference and derive a motion vector of the current block by adding the decoded difference and the median value. This motion vector encoding/decoding scheme may be called a 'median scheme' instead of an 'AMVP scheme'.

In the following embodiments of FIG. 4 and below, a motion prediction process when the AMVP method is used is described, but the present invention is not limited thereto and may be applied in the same or similar manner even when the median method is used.

As another example, when merge is applied, the encoder and the decoder may generate a merge candidate list using motion information of a reconstructed neighboring block and/or motion information of a Col block. That is, the encoder and decoder may use motion information of the reconstructed neighboring block and/or Col block as a merge candidate for the current block, if present.

The encoder may select a merge candidate capable of providing optimal encoding efficiency as motion information for the current block from among merge candidates included in the merge candidate list. At this time, a merge index indicating the selected merge candidate may be included in a bitstream and transmitted to a decoder. The decoder may select one of the merge candidates included in the merge candidate list using the transmitted merge index, and may determine the selected merge candidate as motion information of the current block. Therefore, when merge mode is applied, motion information of the reconstructed neighboring block and/or Col block may be used as motion information of the current block.

In the above-described AMVP and merge modes, motion information of a reconstructed neighboring block and/or motion information of a Col block may be used to derive motion information of a current block. Here, motion information derived from the reconstructed neighboring block may be referred to as spatial motion information, and motion information derived based on the Col block may be referred to as temporal motion information. For example, a motion vector derived from a reconstructed neighboring block may be referred to as a spatial motion vector, and a motion vector derived based on the Col block may be referred to as a temporal motion vector.

Referring back to FIG. 3 , the encoder and decoder may generate a prediction block of the current block by performing motion compensation on the current block based on the derived motion information (S320). Here, the prediction block may refer to a motion-compensated block generated as a result of performing motion compensation on the current block.

4 is a flowchart schematically illustrating an embodiment of a motion estimation process to which the present invention is applied. The motion estimation process according to the embodiment of FIG. 4 may be performed by the motion estimation unit of the image encoding apparatus of FIG. 1 described above.

Referring to FIG. 4 , the encoder may determine a plurality of candidate search points for the current block (S410).

When motion prediction is performed, a search range is determined based on an initial search point, and motion prediction may start at the initial search point. That is, the initial search position is a position at which motion prediction is started during motion estimation, and may mean a position at the center of a search range. Here, the search range may mean a range in which motion prediction is performed within an image and/or a picture.

Accordingly, the encoder may determine a plurality of 'candidate search positions' as candidates for determining an optimal initial search position. Specific embodiments of a method of determining a candidate search position will be described later.

Referring back to FIG. 4 , the encoder may determine a position having the lowest encoding cost as an initial search position among a plurality of candidate search positions (S420).

Here, the encoding cost may mean a cost required to encode the current block. For example, the encoding cost represents a Sum of Absolute Difference (SAD) representing an error value and/or distortion between a current block and a prediction block (where the prediction block may be derived based on a motion vector corresponding to a candidate search position). ), SSE (Sum of Square Error) and/or SSD (Sum of Square Difference) value and the sum of the cost (motion cost) required to encode the motion vector (motion vector corresponding to the candidate search position) may apply to This may be expressed as Equation 1 below, for example.

[Equation 1]

Encoding cost (J) = SAD/SSE/SSD + MV _cost

Here, SAD, SSE, and SSD are error values and/or distortion values between the current block and the prediction block (where the prediction block may be derived based on a motion vector corresponding to a candidate search position) as described above. can represent Specifically, SAD may mean the sum of absolute values of error values between pixel values in an original block and pixel values in a prediction block (where the prediction block may be derived based on a motion vector corresponding to a candidate search position). . In addition, SSE and / or SSD means the sum of squares of error values between pixel values in an original block and pixel values in a prediction block (where the prediction block may be derived based on a motion vector corresponding to a candidate search position) can do. MV _cost may represent a motion cost required to encode a motion vector.

The encoder may generate a prediction block corresponding to the current block for each of a plurality of candidate search positions. The encoder may calculate an encoding cost for each of the generated prediction blocks, and may determine a candidate search location corresponding to a prediction block having the smallest encoding cost as an initial search location.

Referring back to FIG. 4 , the encoder may determine or generate an optimal motion vector for the current block by performing motion estimation based on the determined initial search position (S430).

As described above, the encoder may set a search range based on an initial search position. Here, the initial search location may be located at the center of the search range, and the size and/or shape of the search range may be determined as a predetermined size and/or shape. At this time, the encoder may determine the location of the pixel having the minimum error value (or minimum encoding cost) by performing motion prediction within the set search range. Here, the position of the pixel having the minimum error value may mean a position indicated by an optimal motion vector generated by performing motion prediction on the current block. That is, the encoder may determine a motion vector indicating a position of a pixel having a minimum error value (or minimum coding cost) as a motion vector of the current block.

For example, the encoder may generate a plurality of prediction blocks centered on positions of pixels existing within a set search range. In this case, the encoder may determine an encoding cost corresponding to each position of pixels existing within a search range based on the plurality of prediction blocks and the original block. Also, the encoder may determine a motion vector corresponding to a position of a pixel having the smallest coding cost as a motion vector of the current block.

Also, when motion prediction is performed on all pixels within a search range, complexity may be excessively high. Accordingly, the encoder may perform a pattern search for performing motion prediction based on only pixels indicated by a predetermined pattern within a set search range.

When the motion vector of the current block is derived or generated, the encoder can generate a prediction block corresponding to the current block by performing motion compensation on the current block based on it. The encoder may generate a residual block based on a difference between the current block and the predicted block, and may output a bitstream by performing transformation, quantization, and/or entropy encoding on the generated residual block.

According to the above-described embodiment, it may be determined whether a pixel having a minimum error value is included in a search range according to where an initial search position is determined. In addition, as the correlation between the initial search position and the position of the pixel with the minimum error value increases, the encoder can more efficiently find the position of the pixel with the minimum error value in motion estimation. Accordingly, various methods of determining an initial search position may be provided in order to improve coding efficiency and reduce complexity of motion prediction.

5 is a diagram schematically illustrating an embodiment of a method for determining an initial search location.

5 illustrates a current picture 510 to which the current block (BLK _Current ) belongs and a reference picture 520 used for inter prediction of the current block (BLK _Current ). In the embodiment of FIG. 5 , BLK _B and BLK _C may indicate neighboring blocks located adjacent to the current block.

Referring to 510 of FIG. 5 , the encoder may determine a plurality of candidate search positions for the current block based on motion vectors of neighboring blocks adjacent to the current block.

For example, the encoder may determine the position 513 indicated by the predicted motion vector (MV _PMV ) of the current block as the candidate search position of the current block based on the position of the zero point 516 . As described above, the predictive motion vector may be determined through the AMVP method or the median method. For example, when the AMVP scheme is applied, the predicted motion vector (MV _PMV ) of the current block may be derived based on the motion vector of the reconstructed neighboring block and/or the motion vector of the Col block. Accordingly, the number of predictive motion vectors of the current block may be plural. In the embodiment of FIG. 5 , for convenience of description, only the candidate search position 513 indicated by one predictive motion vector (MV _PMV ) is shown, but the present invention is not limited thereto and all of the plurality of predictive motion vectors used in AMVP are shown. It can be used to determine the candidate search position of the current block.

Also, the encoder may determine a zero point 516 located at the center of the current block (BLK _Current ) as a candidate search position of the current block. Here, the zero point position 516 may be represented by a zero vector (MV _Zero ), and for example, the zero vector (MV _Zero ) may be (0,0).

Also, the encoder may determine a position indicated by a motion vector of a neighboring block located adjacent to the current block as a candidate search position of the current block based on the zero point position 516 . For example, the encoder may determine a position 519 indicated by a motion vector (MV _B ) of a leftmost block (BLK _B ) among blocks adjacent to an upper end of the current block as a candidate search position of the current block. In the embodiment of FIG. 5 , only the position 519 indicated by the motion vector of the block BLK _B among blocks located around the current block (BLK _Current ) is shown as a candidate search position, but the present invention is not limited thereto. For example, the encoder may determine a position indicated by a motion vector of a block adjacent to the left of the current block (BLK _Current ) as a candidate search position, and may determine the motion of the block (BLK _C ) located at the upper right corner outside the current block (BLK _Current ). A position indicated by the vector may be determined as a candidate search position.

When a plurality of candidate search positions are determined, the encoder may generate a prediction block corresponding to the current block for each of the plurality of candidate search positions 513, 516, and 519. Also, the encoder may calculate an encoding cost for each generated prediction block. At this time, the encoder may determine a candidate search position corresponding to a prediction block having the smallest coding cost as an initial search position among a plurality of candidate search positions 513 , 516 , and 519 . Since the embodiment of the encoding cost calculation method has been described in detail with reference to FIG. 4, a detailed description thereof will be omitted here.

Referring to 520 of FIG. 5 , for example, an initial search position 513 may be determined as a position 513 indicated by a predicted motion vector (MV _PMV ) of the current block. At this time, the encoder may generate an optimal motion vector for the current block by performing motion prediction based on the determined initial search position 513 .

As described above, the encoder may set the search range 525 based on the initial search position 513 . Here, the initial search position 513 may be located at the center of the search range 525, and the size and/or shape of the search range 525 may be set to a predetermined size and/or shape. At this time, the encoder may determine the position of the pixel having the minimum error value (or minimum encoding cost) by performing motion prediction within the set search range 525 . The encoder may determine the motion vector representing the determined position as the motion vector of the current block.

According to the above-described embodiment, in determining an initial search position, the encoder may refer to a motion vector value of a neighboring block having a value similar to that of the current block. Motion vectors of neighboring blocks adjacent to the current block may be similar to those of the current block in many cases. However, in a case where the number of block divisions is increased due to a complicated motion and/or texture in the current block, correlation between motion vectors of the current block and motion vectors of neighboring blocks may be low.

If the initial search position is determined by referring to the motion vectors of the neighboring blocks when the correlation between the motion vectors of the current block and the motion vectors of the neighboring blocks is low, there is a high possibility that a pixel having a minimum error value is not included in the search range. In addition, in this case, since the distance between the initial search position and the position of the pixel with the minimum error value is likely to be long, motion prediction must be performed at more pixel positions to find the pixel with the minimum error value during pattern search. may be

Therefore, in order to improve coding efficiency and reduce complexity of motion prediction, various methods of determining an initial search position may be provided in addition to a method of determining an initial search position by referring to motion vectors of neighboring blocks adjacent to the current block.

6 is a diagram schematically illustrating an embodiment of a method for determining a candidate search position according to the present invention.

In the embodiment of FIG. 6 , a dotted line arrow may mean a motion vector derived by motion prediction, and a solid line arrow indicates a motion vector (eg, predicted motion vector) indicating a candidate search position determined according to the embodiment of FIG. 5 . can mean

In the process of encoding an image, an encoding target block may be further divided into smaller sub-blocks. In this case, the encoder may perform motion prediction on each of the divided sub-blocks after performing motion prediction on the encoding target block before being divided.

When the current block is a sub-block generated by dividing the current block and motion prediction of the current block has already been completed, the encoder selects a position indicated by a motion vector derived from the motion prediction of the current block as a candidate search position. can be determined by

In addition, sub-blocks generated by dividing the encoding object block may be plural. Therefore, before motion prediction of the current block corresponding to the sub-block is performed, a sub-block for which motion prediction has already been completed may exist in the encoding target block, which may be a neighboring block located adjacent to the current block in the encoding target block. there is. At this time, the encoder may determine a position indicated by a motion vector of the sub-block for which the motion prediction is completed as a candidate search position.

Hereinafter, in this specification, for convenience of description, the encoding object block including the lower block is referred to as an upper block for a lower block generated by dividing the encoding target block. For example, when the current block is a lower block generated by splitting the encoding object block, the encoding object block including the current block may be regarded as an upper block of the current block. Since the lower block is a block created by dividing the upper block, the size of the upper block may be greater than that of the lower block.

In 610 of FIG. 6 , BLK _64x64 indicates an uppermost block, and for example, the uppermost block may have a size of 64x64. Since there is no upper block for the uppermost block (BLK _64x64 ), a motion vector that can be used to determine a candidate search position may not exist in the uppermost block when motion prediction is performed on the uppermost block. That is, since motion prediction has not yet been performed on the uppermost block, a motion vector that can be used in the uppermost block may not exist.

Accordingly, the encoder may determine the candidate search position by the method described above in the embodiment of FIG. 5 .

For example, the encoder may determine the zero position 613 located in the center of the uppermost block (BLK _64x64 ) as the candidate search position of the uppermost block (BLK _64x64 ). Here, the zero point (613) may be represented by a zero vector, and for example, the zero vector may be (0,0). In addition, the encoder may determine a position 616 indicated by a predicted motion vector (MV _AMVP ) of the highest block (BLK _64x64 ) as a candidate search position of the highest block (BLK _64x64 ) based on the zero point position 613 .

In 610 of FIG. 6 , for convenience of description, only the position 613 indicated by the zero vector and the position 616 indicated by the predicted motion vector (MV _AMVP ) are shown as candidate search positions, but the present invention is not limited thereto. For example, as in the embodiment of FIG. 5 , the encoder may determine a position indicated by a motion vector of a neighboring block adjacent to the uppermost block (BLK _64x64 ) as a candidate search position.

620 of FIG. 6 shows an uppermost block (BLK _64x64 ) on which motion prediction is performed. In 620 of FIG. 6 , MV _64x64 may indicate a motion vector generated by motion prediction for an uppermost block (BLK _64x64 ), and MV _64x64 may indicate a position 623 within the uppermost block (BLK _64x64 ).

Referring to 630 of FIG. 6 , an uppermost block (BLK _64x64 ) may be divided into a plurality of lower blocks (BLK ¹ _32x32 , BLK ² _32x32 , BLK ³ _32x32 , and BLK ⁴ _32x32 ). At this time, in one embodiment, each of the plurality of sub-blocks may have a size of 32x32. For example, the lower block BLK ¹ _32x32 may be a block located at the upper left of the uppermost block (BLK _64x64 ), and the lower block BLK ² _32x32 may be a block located on the upper right within the uppermost block (BLK _64x64 ). Also, the lower block BLK ³ _32x32 may be a block positioned at the lower left of the uppermost block (BLK _64x64 ), and the lower block BLK ⁴ _32x32 may be a block located on the lower right of the uppermost block (BLK _64x64 ).

After performing motion prediction on the uppermost block (BLK _64x64 ), the encoder may also perform motion prediction on each lower block (BLK ¹ _32x32 , BLK ² _32x32 , BLK ³ _32x32 , BLK ⁴ _32x32 ). At this time, the encoder may perform motion prediction on the lower block in the order of, for example, BLK ¹ _32x32 , BLK ² _32x32 , BLK ³ _32x32 , and BLK ⁴ _32x32 .

Referring again to 630 of FIG. 6 , the encoder may perform motion prediction on a first block (BLK ¹ _32x32 ) among sub-blocks. At this time, the encoder selects the candidate search positions 633 and 636 derived according to the embodiment of FIG. 5 and the position 639 indicated by the motion vector (MV _64x64 ) generated by motion prediction of the upper block (BLK _64x64 ). At least one may be determined as a candidate search position.

For example, the encoder may determine the zero position 633 located in the center of the first block (BLK ¹ _32x32 ) as a candidate search position. Here, the zero point position 633 may be represented by a zero vector. In addition, the encoder may determine a position 636 indicated by a predicted motion vector (MV _AMVP ) of the first block (BLK ¹ _32x32 ) based on the zero point position 633 as a candidate search position. Also, the encoder may determine a position 639 indicated by a motion vector (MV _64x64 ) generated by motion prediction of an uppermost block (BLK _64x64 ) as a candidate search position.

630 of FIG. 6 shows only positions 633, 636, and 639 as candidate search positions within the first block (BLK ¹ _32x32 ), but the present invention is not limited thereto. For example, the encoder may additionally determine a position indicated by a motion vector of a neighboring block located adjacent to the first block (BLK ¹ _32x32 ) as a candidate search position.

640 of FIG. 6 is an embodiment of a method for determining a candidate search position for a second block (BLK ² _32x32 ) when motion prediction is completed for the first block (BLK ¹ _32x32 , 645) in the uppermost block (BLK _64x64 ). indicates In 640 of FIG. 6 , MV ¹ _32x32 may indicate a motion vector generated by motion prediction for the first block (BLK ¹ _32x32 , 645). MV ¹ _32x32 may indicate position 653 within the first block (BLK ¹ _32x32 , 645).

Referring to 640 of FIG. 6 , the encoder may perform motion prediction on a second block (BLK ² _32x32 ) among sub-blocks. At this time, the encoder selects candidate search positions 662 and 664 derived according to the embodiment of FIG. 5, a position 666 indicated by a motion vector (MV _64x64 ) generated by motion prediction for an upper block (BLK _64x64 ), At least one of positions 639 indicated by motion vectors (MV ¹ _32x32 ) of other lower blocks (BLK ¹ _32x32 , ₆₄₅ ) in the upper block (BLK 64x64) for which motion prediction has already been completed may be determined as a candidate search position. Here, another sub-block (BLK ¹ _32x32 , 645) for which motion prediction has already been completed may be a block located around a sub-block (BLK ² _32x32 ) to be motion-predicted within an upper block (BLK _64x64 ).

For example, the encoder may determine the zero position 662 located in the center of the second block (BLK ² _32x32 ) as a candidate search position. Here, the zero point location 662 may be represented by a zero vector. In addition, the encoder may determine a position 664 indicated by a predicted motion vector (MV _AMVP ) of the second block (BLK ² _32x32 ) based on the zero point position 662 as a candidate search position.

Also, the encoder may determine a position 639 indicated by a motion vector (MV _64x64 ) generated by motion prediction of an uppermost block (BLK _64x64 ) as a candidate search position. Also, the encoder may determine a position ₆₆₈ indicated by a motion vector (MV ¹ _32x32 ) of a sub-block (BLK ¹ _32x32 , 645) for which motion prediction has already been completed among sub-blocks in the uppermost block (BLK 64x64 ) as a candidate search position. there is.

For each of the remaining sub-blocks (BLK ³ _32x32 , BLK ⁴ _32x32 ), the encoder may determine a candidate search position in a manner similar to that of the second block (BLK ² _32x32 ). For example, for each of the lower blocks BLK ³ _32x32 and BLK ⁴ _32x32 , the encoder generates a motion vector (MV _64x64 ) generated by motion prediction of the candidate search position derived according to the embodiment of FIG. 5 and the upper block (BLK _64x64 ). At least one of a position indicated by and a position indicated by a motion vector of another sub-block for which motion prediction has already been completed within the upper block (BLK _64x64 ) may be determined as a candidate search position.

640 of FIG. 6 shows only positions 662, 664, 666, and 668 as candidate search positions within the second block (BLK ² _32x32 ), but the present invention is not limited thereto. For example, the encoder may additionally determine a position indicated by a motion vector of a neighboring block located adjacent to the second block (BLK ² _32x32 ) as a candidate search position.

7 is a diagram schematically illustrating another embodiment of a method for determining a candidate search position according to the present invention.

7 shows a current picture (current picture, 710) to which a current block (BLK _Current ), which is a motion prediction target, and a reference picture (reference picture, 720) used for inter prediction of the current block (BLK _Current ) are shown in FIG. . Here, the reference picture 720 may be a picture that has already been coded and/or decoded, and includes blocks (BLK _Colocated , BLK _A , BLK _B , BLK _C , BLK _D , BLK _E , BLK) belonging to the reference picture 720. _F ) may also be blocks that have all been coded and/or decoded. In the embodiment of FIG. 7 , the motion vector for BLK _Colocated is MV _Colocated , and the motion vectors for BLK _A , BLK _B , BLK _C , BLK _D , BLK _E and BLK _F are MV _A , MV _B , MV _C , They are referred to as MV _D , MV _E and MV _F .

When performing motion prediction, the encoder may determine positions indicated by motion vectors of blocks belonging to the reference picture 720 as candidate search positions of the current block (BLK _Current ).

For example, the encoder locates a position indicated by a motion vector (MV _Colocated ) of a block (BLK _Colocated ) existing at the same spatial position (overlapping position) as the current block (BLK _Current ) in the reference picture 720. _Current ) can be determined as a candidate search position. Here, a block (BLK _Colocated ) existing at the same spatial position (overlapping position) as the current block (BLK _Current ) in the reference picture 720 may be referred to as a 'collocated block'.

In addition, the encoder uses motion vectors MV _A of neighboring blocks BLK _A , BLK _B , BLK _C , BLK _D , BLK _E , and BLK _F located adjacent to the colocated block BLK _Colocated in the reference picture 720 . , MV _B , MV _C , MV _D , MV _E , and MV _F ) may determine at least one of locations indicated by the candidate search location of the current block (BLK _Current ). In FIG. 7, since the blocks in the reference picture 720 are already coded and/or decoded blocks, the motion vector of not only the co-located block but also all blocks adjacent to the co-located block is a candidate search position of the current block. can be used to determine

8 is a diagram schematically illustrating another embodiment of a method for determining a candidate search position according to the present invention.

In the embodiment of FIG. 8 , a dotted line arrow may mean a motion vector derived by motion prediction, and a solid line arrow indicates a motion vector (eg, predicted motion vector) indicating a candidate search position determined according to the embodiment of FIG. 5 . can mean

8 shows an upper block BLK _64x64 , lower blocks generated by dividing the upper block BLK ¹ _32x32 , BLK ² _32x32 , BLK ³ _32x32 , BLK ⁴ _32x32 , and blocks BLK _A and BLK _B located adjacent to the upper block. , BLK _C is shown. Here, as an example, the size of the upper block (BLK _64x64 ) may be 64x64, and the size of the lower blocks (BLK ¹ _32x32 , BLK ² _32x32 , BLK ³ _32x32 , BLK ⁴ _32x32 ) may be 32x32, respectively. For example, motion prediction for lower blocks may be performed in the order of BLK ¹ _32x32 , BLK ² _32x32 , BLK ³ _32x32 , and BLK ⁴ _32x32 .

8, MV _64x64 represents a motion vector generated by motion prediction for an upper block (BLK _64x64 ), and MV ¹ _32x32 represents a motion vector generated by motion prediction for a first lower block (BLK ¹ _32x32 ). can represent Also, MV _A , MV _B , and MV _C may represent motion vectors generated by motion prediction of neighboring blocks BLK _A , BLK _B , and BLK _C , respectively. And, MV _AMVP may indicate a predicted motion vector.

As described above with reference to FIGS. 5 to 7 , the encoder may determine a candidate search position for a motion prediction target block using various methods. For example, it is assumed that the current block is a lower block generated by dividing an upper block. At this time, the encoder uses a zero point position (hereinafter, a motion vector representing the zero point position is referred to as a first motion vector), a position indicated by the predicted motion vector (hereinafter referred to as a second motion vector), and a neighboring block adjacent to the motion prediction target block. A position indicated by a motion vector of (hereinafter referred to as a third motion vector), a position indicated by a motion vector (hereinafter referred to as a fourth motion vector) of an upper block for a motion prediction target block, and among lower blocks within the upper block The position indicated by the motion vector (hereinafter referred to as the fifth motion vector) of the block for which motion prediction has already been completed, the position indicated by the motion vector (hereinafter referred to as the sixth motion vector) of the block at the same position in the reference picture, and the position within the reference picture At least one of positions indicated by a motion vector (hereinafter, referred to as a seventh motion vector) of a block adjacent to the co-located block may be determined as a candidate search position for the motion prediction target block.

The above-described first to seventh motion vectors may constitute a set of motion vectors usable for motion prediction of the current block. Hereinafter, in this specification, for convenience of explanation, a set of motion vectors usable for motion prediction of a current block will be referred to as a 'motion vector set'.

The encoder may generate a new motion vector by combining at least one motion vector among a plurality of motion vectors constituting the motion vector set. For example, the encoder may use a value generated by adding an average value, a maximum value, a minimum value, and/or a weight of one or more motion vectors included in the motion vector set as a new motion vector value. At this time, the encoder may determine a position indicated by the new motion vector as a candidate search position.

In FIG. 8, it is assumed that an encoder performs motion prediction on a second block (BLK ² _32x32 ) among lower blocks in an upper block (BLK _64x64 ). That is, in the embodiment of FIG. 8 , the current block to be motion prediction may be a second block (BLK ² _32x32 ) among lower blocks in an upper block (BLK _64x64 ).

At this time, in FIG. 8, the upper block (BLK _64x64 ), blocks adjacent to the upper block (BLK _A , BLK _B , BLK _C ), and the first lower block (BLK ¹ _32x32 ) have already performed motion prediction. This may be a completed block. In this case, each block on which motion prediction is completed may include a motion vector generated by motion prediction.

Therefore, in the embodiment of FIG. 8 , for example, a motion vector set, which is a set of motion vectors usable for motion prediction of a current block (BLK ² _32x32 ), is a motion vector of an upper block (MV _64x64 ), a neighbor adjacent to the upper block may include motion vectors (MV _A , MV _B , and MV _C ) of neighboring blocks located in , a motion vector (MV ¹ _32x32 ) of a first sub-block, and a predicted motion vector (MV _AMVP ). 8, MV _64x64 is (-6,6), MV ¹ _32x32 is (-5,2), MV _AMVP is (8,-2), MV _A is (0,10), and MV _B is ( -3,10) and MV _C is assumed to be (6,0).

In this case, the encoder may generate a new motion vector by combining one or more motion vectors from among a plurality of motion vectors included in the motion vector set. In the embodiment of FIG. 8 , among a plurality of motion vectors included in a motion vector set, the motion vectors used to generate a new motion vector are the motion vector of the upper block (MV _64x64 ) and the motion vector of the first lower block (MV ¹ _32x32 ), and a predictive motion vector (MV _AMVP ).

For example, the encoder may determine an average value of motion vectors as a new motion vector value. In this case, a new motion vector can be calculated by Equation 2 below.

[Equation 2]

X = (8-6-5)/3 = -1, Y = (-2+6+2)/3 = 2

MV _MEAN = (X,Y) = (-1,2)

Here, MV _MEAN may represent a new motion vector derived by averaging motion vectors included in the motion vectors.

As another example, the encoder may determine the maximum value among the X components of the motion vectors as the value of the X component of the new motion vector, and determine the maximum value among the Y components of the motion vectors as the value of the Y component of the new motion vector. In this case, a new motion vector can be calculated by Equation 3 below.

[Equation 3]

X=8, Y=6, MV _MAX =(8,6)

Here, MV _MAX may represent a motion vector newly derived by the above method.

As another example, the encoder may determine the minimum value among the X components of the motion vectors as the value of the X component of the new motion vector, and determine the minimum value among the Y components of the motion vectors as the value of the Y component of the new motion vector. In this case, a new motion vector can be calculated by Equation 4 below.

[Equation 4]

X=-6, Y=-2, MV _MIN =(-6,-2)

Here, MV _MIN may represent a motion vector newly derived by the above method.

When a new motion vector is generated by combining one or more motion vectors among a plurality of motion vectors included in the motion vector set, the encoder may determine a position indicated by the generated motion vector as a candidate search position of the current block (BLK ² _32x32 ). there is.

Meanwhile, when a candidate search position is determined as in the above-described embodiments, the encoder may determine a position having the lowest coding cost among a plurality of candidate search positions as an initial search position, as described above in the embodiment of FIG. 4 . there is. In this case, for example, the encoder may determine an initial search position by calculating an encoding cost for all candidate search positions derived for the motion prediction target block. However, the method of calculating the encoding cost for all candidate search positions may be complex. Accordingly, the encoder may determine the initial search position after reducing the number of candidate search positions based on the degree of correlation between motion vectors in the process of determining the initial search position.

As an embodiment, the encoder may remove a position indicated by a motion vector having the largest difference with the predicted motion vector (PMV) from among a plurality of candidate search positions derived for the current block. In addition, the encoder selects a predetermined number (eg, 2, 3, or 4, etc.) in order of greatest difference from the predicted motion vector (PMV) among motion vectors representing a plurality of candidate search positions derived for the current block. It is also possible to select a motion vector of and remove a position indicated by the selected motion vector. Here, the difference value between the motion vectors may correspond to, for example, a value obtained by adding the absolute value of the difference between the X components and the absolute value of the difference between the Y components.

In another embodiment, the encoder searches only a position indicated by a motion vector having the smallest difference with the predicted motion vector (PMV) and a position indicated by the predicted motion vector (PMV) among a plurality of candidate search positions derived for the current block. position can be used. That is, in this case, the encoder can remove all positions other than the position indicated by the motion vector having the smallest difference value from the predicted motion vector (PMV) and the position indicated by the predicted motion vector (PMV). In addition, the encoder selects a predetermined number (e.g., 2, 3, or 4, etc.) in order of small differences from the predicted motion vector (PMV), among motion vectors representing a plurality of candidate search positions derived for the current block. A motion vector may be selected, and only the position indicated by the selected motion vector and the position indicated by the predicted motion vector (PMV) may be used as candidate search positions. That is, in this case, the encoder can remove all positions other than the positions indicated by the predetermined number of motion vectors and the predicted motion vectors (PMV).

For example, it is assumed that positions indicated by MV _64x64 , MV ¹ _32x32 , MV _AMVP , MV _A , MV _B and MV _C in FIG. 8 are determined as candidate search positions of the current block (MV ² _32x32 ). Here, for example, MV _64x64 is (-6,6), MV ¹ _32x32 is (-5,2), MV _AMVP is (8,-2), MV _A is (0,10), MV _B is (-3 ,10), MV _C may be (6,0). In this case, a difference value between each motion vector representing the candidate search position and the predicted motion vector (MV _AMVP ) may be calculated as in Equation 5 below.

[Equation 5]

|MV _AMVP - MV _64x64 | = |{8-(-6)}| + |(-2-6)| = 22

|MV _AMVP - MV ¹ _32x32 | = |{8-(-5)}| + |(-2-2)| = 17

|MV _AMVP - MV _A | = |8-0| + |(-2-10)| = 20

|MV _AMVP - MV _B | = |{8-(-3)}| + |(-2-10)| = 23

|MV _AMVP - MV _C | = |8-6| + |(-2-0)| = 4

In this case, the encoder may remove, for example, a position indicated by a motion vector (MV _B ) having the largest difference with the predicted motion vector from the candidate search position. As another example, the encoder searches candidates for a position indicated by a motion vector (MV _B ) having the largest difference from the predicted motion vector and a position indicated by a motion vector (MV _64x64 ) having the next greatest difference from the predicted motion vector (MV _B ). can be removed from position. Here, the difference value between the motion vectors may correspond to, for example, a value obtained by adding the absolute value of the difference between the X components and the absolute value of the difference between the Y components.

As another example, the encoder may use only positions indicated by a predicted motion vector (MV _AMVP ) and a motion vector (MV _C ) having the smallest difference between the predicted motion vector and the predicted motion vector, as candidate search positions. In this case, the encoder may remove positions other than positions indicated by the motion vectors MV _AMVP and MV _C from the candidate search position. As another example, the encoder determines a predicted motion vector (MV _AMVP ), a position indicated by a motion vector (MV _C ) having the smallest difference value from the predicted motion vector, and a motion having a difference value next to MV _C that is smaller than the predicted motion vector. Only the position indicated by the vector (MV ¹ _32x32 ) may be used as the candidate search position. In this case, the encoder may remove positions other than positions indicated by the motion vectors MV _AMVP , MV _C , and MV ¹ _32x32 from the candidate search position.

As another embodiment, the encoder may remove a position indicated by a motion vector having the greatest difference from a motion vector of an upper block, among a plurality of candidate search positions derived for the current block. In addition, the encoder selects a predetermined number (eg, 2, 3, or 4, etc.) in order of greatest difference from the motion vector of the upper block, among motion vectors representing a plurality of candidate search positions derived for the current block. It is also possible to select a motion vector of and remove a position indicated by the selected motion vector.

In another embodiment, the encoder selects only a position indicated by a motion vector having the smallest difference value with respect to a motion vector of an upper block and a position indicated by the motion vector of an upper block among a plurality of candidate search positions derived for the current block. It can be used as a candidate search location. That is, in this case, the encoder can remove all positions other than the position indicated by the motion vector having the smallest difference with the motion vector of the upper block and the position indicated by the motion vector of the upper block. In addition, the encoder selects a predetermined number (eg, 2, 3, or 4) of motion vectors in order of small difference values from motion vectors of upper blocks among a plurality of candidate search positions derived for the current block. can be selected, and only the position indicated by the selected motion vector and the position indicated by the motion vector of the upper block may be used as the candidate search position. That is, in this case, the encoder can remove all positions other than the positions indicated by the predetermined number of motion vectors and the motion vectors of the upper block.

For example, it is assumed that positions indicated by MV _64x64 , MV ¹ _32x32 , MV _AMVP , MV _A , MV _B and MV _C in FIG. 8 are determined as candidate search positions of the current block (MV ² _32x32 ). Here, for example, MV _64x64 is (-6,6), MV ¹ _32x32 is (-5,2), MV _AMVP is (8,-2), MV _A is (0,10), MV _B is (-3 ,10), MV _C may be (6,0). In this case, a difference value between each motion vector representing the candidate search position and the motion vector (MV _64x64 ) of the upper block may be calculated as in Equation 6 below.

[Equation 6]

|MV _64x64 - MV _AMVP | = |-6-8| + |(6-(-2)}| = 22

|MV _64x64 - MV ¹ _32x32 | = |{-6-(-5)}| + |(6-2)| = 5

|MV _64x64 - MV _A | = |-6-0| + |6-10| = 10

|MV _64x64 - MV _B | = |{-6-(-3)}| + |6-10| = 7

|MV _64x64 - MV _C | = |-6-6| + |6-0| = 18

In this case, the encoder may, for example, remove a position indicated by a motion vector (MV _AMVP ) having the largest difference with the motion vector of the upper block from the candidate search position. As another example, the encoder determines the position indicated by the motion vector (MV _AMVP ) having the largest difference value from the motion vector of the upper block, and the motion vector (MV _C ) having the next largest difference value from the motion vector of the upper block (MV _AMVP ). A location can be removed from the list of candidate search locations.

As another example, the encoder can use only the positions indicated by the motion vector (MV _64x64 ) of the upper block and the motion vector (MV ¹ _32x32 ) having the smallest difference between the motion vector and the upper block motion vector, as candidate search positions. In this case, the encoder may remove positions other than positions indicated by the motion vectors MV _64x64 and MV ¹ _32x32 from the candidate search positions. As another example, the encoder may use a motion vector (MV _64x64 ) of an upper block, a position indicated by a motion vector (MV ¹ _32x32 ) having the smallest difference value with the motion vector of the upper block, and a difference value with the motion vector of the upper block. Only positions indicated by motion vectors (MV _B ) next to MV ¹ _32x32 may be used as candidate search positions. In this case, the encoder may remove positions other than positions indicated by motion vectors MV _64x64 , MV ¹ _32x32 , and MV _B from the candidate search position.

As another embodiment, when the current block (eg, MV ² _32x32 in FIG. 8 ) is a lower block generated by dividing an upper block (eg, MV _64x64 in FIG. 8 ), motion prediction has already been performed in the upper block. Another sub-block (eg, MV ¹ _32x32 in FIG. 8 ) may exist. Hereinafter, 'another lower block' may refer to a lower block on which motion prediction has already been performed while belonging to the same upper block as the current block.

In this case, as an example, the encoder selects a position indicated by a motion vector having the greatest difference from a motion vector of the other sub-block (eg, MV ¹ _32x32 in FIG. 8 ) among a plurality of candidate search positions derived for the current block. can be removed In addition, the encoder selects a predetermined number of motion vectors (eg, MV ¹ _32x32 in FIG. 8 ) in order of greatest difference from motion vectors of other sub-blocks (eg, MV 1 32x32 in FIG. 8 ) among motion vectors indicating a plurality of candidate search positions derived for the current block. , 2, 3, 4, etc.) motion vectors may be selected and positions indicated by the selected motion vectors may be removed.

In this case, as another example, the encoder selects a position indicated by a motion vector having the smallest difference value from a motion vector of another sub-block (eg, MV ¹ _32x32 in FIG. 8 ) among a plurality of candidate search positions derived for the current block, and Only positions indicated by motion vectors of other sub-blocks (eg, MV ¹ _32x32 in FIG. 8 ) may be used as candidate search positions. That is, in this case, the encoder determines the position indicated by the motion vector having the smallest difference value with the motion vector of the other sub-block (eg, MV ¹ _32x32 in FIG. 8 ) and the motion of the other sub-block (eg, MV ¹ _32x32 in FIG. 8 ). All positions other than the position indicated by the vector can be removed. In addition, _the encoder selects a predetermined number ( ^e.g. , 2, 3 or 4 motion vectors may be selected, and only positions indicated by the selected motion vectors and motion vectors of other sub-blocks (eg, MV ¹ _32x32 in FIG. 8 ) may be used as candidate search positions. . That is, in this case, the encoder can remove all positions other than the positions indicated by the predetermined number of motion vectors and the motion vectors of the other sub-blocks (eg, MV ¹ _32x32 in FIG. 8 ).

Since a specific embodiment of a method for determining a position to be removed from a candidate search position based on a motion vector of another sub-block is similar to Equations 5 and 6, a detailed description thereof will be omitted here.

As another embodiment, the encoder may obtain a variance value for each of the motion vectors based on motion vectors indicating a plurality of candidate search positions derived for the current block. At this time, the encoder may determine a position to be removed from the candidate search position based on the variance value.

For example, the encoder may remove a position indicated by a motion vector having the largest variance value among a plurality of candidate search positions derived for the current block. In addition, the encoder selects a predetermined number (eg, 2, 3, or 4) of motion vectors in order of increasing variance values among motion vectors representing a plurality of candidate search positions derived for the current block, and A position indicated by the selected motion vector may be removed.

As another example, the encoder may use only a position indicated by a motion vector having the smallest variance value as a candidate search position among a plurality of candidate search positions derived for the current block. That is, in this case, the encoder can remove all positions other than the position indicated by the motion vector having the smallest variance value. In addition, the encoder may select a predetermined number (eg, 2, 3, or 4) of motion vectors in descending order of variance values from motion vectors representing a plurality of candidate search positions derived for the current block, , only positions indicated by the selected motion vectors may be used as candidate search positions. That is, in this case, the encoder can remove all positions other than the positions indicated by the predetermined number of motion vectors.

Since a specific embodiment of a method for determining a position to be removed from a candidate search position based on a variance value is similar to Equations 5 and 6 described above, a detailed description thereof will be omitted here.

When a position to be removed from among a plurality of candidate search positions derived for the current block is determined according to the above-described embodiments, the encoder may determine an optimal initial search position among the remaining candidate search positions excluding the removed position. At this time, for example, the encoder may determine, as an initial search position, a position having a minimum encoding cost among the remaining candidate search positions excluding the removed position.

According to the above-described embodiments, the encoder may refer to a motion vector of a block having a high correlation with the current block when performing motion prediction of the current block. In particular, since an upper block to which the current block belongs and other lower blocks belonging to the upper block have a high correlation with the current block, the encoder can find the position of the pixel having the minimum error value more efficiently.

In addition, when the process of determining the candidate search position and the process of determining the initial search position are performed according to the above-described embodiments, the probability that the position of the pixel having the minimum error value is included in the search range may increase. Also, according to the above-described embodiments, in a motion estimation process such as pattern search, the encoder can more quickly find the position of a pixel having a minimum error value. Therefore, according to the present invention, encoding performance can be improved.

In the foregoing embodiments, the methods are described on the basis of a flowchart as a series of steps or blocks, but the present invention is not limited to the order of steps, and some steps may occur in a different order or concurrently with other steps as described above. can In addition, those skilled in the art will understand that the steps shown in the flow chart are not exclusive, that other steps may be included, or that one or more steps of the flow chart may be deleted without affecting the scope of the present invention. You will understand.

The foregoing embodiment includes examples of various aspects. It is not possible to describe all possible combinations to represent the various aspects, but those skilled in the art will recognize that other combinations are possible. Accordingly, it is intended that the present invention cover all other substitutions, modifications and variations falling within the scope of the following claims.

Claims (1)

selecting an initial search position for the current block;
deriving a motion vector of the current block by comparing encoding costs within a set search range based on the initial search position;
generating a prediction block by performing prediction on the current block based on the derived motion vector;
generating a reconstruction block based on the prediction block; and
Including filtering the reconstruction block,
The form of the search range is a predetermined form,
The step of deriving the motion vector of the current block,
Performing a pattern search based on pixels indicated by a predetermined pattern within the search range;
The encoding cost is calculated using SAD (Sum of Absolute Difference),
The inter prediction method, characterized in that the filtering comprises deblocking filtering.

Images