KR20210131299A - Method for inter prediction and apparatus thereof - Google Patents

Abstract

An image decoding apparatus according to the present invention comprises: a motion prediction part for deriving motion information of a current block as motion information including L0 motion information and L1 motion information; a motion compensation part that generates a prediction block corresponding to the current block by performing motion compensation on the current block based on at least one of the L0 motion information and the L1 motion information; and a reconstructed block generation part that generates a reconstructed block corresponding to the current block based on the prediction block. According to the present invention, image encoding efficiency can be improved.

Description

Inter prediction method and device therefor

The present invention relates to image processing, and more particularly, to an inter prediction method and an apparatus therefor.

Recently, as broadcasting services with high definition (HD) resolution are expanding not only domestically but also worldwide, many users are getting used to high-resolution and high-definition images, and accordingly, many institutions are spurring the development of next-generation imaging devices. In addition, as interest in UHD (Ultra High Definition) having a resolution four times higher than that of HDTV increases along with HDTV, compression technology for higher resolution and high-definition images is required.

For image compression, inter prediction technology that predicts pixel values included in the current picture from temporally previous and/or later pictures, predicting pixel values included in the current picture using pixel information in the current picture An intra prediction technique, an entropy encoding technique in which a short code is assigned to a symbol having a high frequency of occurrence and a long code is assigned to a symbol having a low frequency of occurrence, may be used.

An object of the present invention is to provide an image encoding method and apparatus capable of improving image encoding efficiency.

Another technical object of the present invention is to provide an image decoding method and apparatus capable of improving image encoding efficiency.

Another technical object of the present invention is to provide an inter prediction method and apparatus for improving image encoding efficiency.

Another technical object of the present invention is to provide a method and apparatus for deriving motion information capable of improving image encoding efficiency.

Another technical object of the present invention is to provide a method and apparatus for deriving temporal motion information capable of improving image encoding efficiency.

One embodiment of the present invention is an image decoding apparatus. The apparatus includes a motion predictor for deriving motion information of a current block as motion information including L0 motion information and L1 motion information, and a motion with respect to the current block based on at least one of the L0 motion information and the L1 motion information. A motion compensation unit generating a prediction block corresponding to the current block by performing compensation and a reconstructed block generation unit generating a reconstructed block corresponding to the current block based on the prediction block, wherein the motion compensator comprises: The motion compensation is performed based on whether the L0 motion information and the L1 motion information are identical.

The L0 motion information includes an L0 motion vector and an L0 reference picture number, the L0 reference picture number indicates a picture order count (POC) assigned to an L0 reference picture, and the L1 motion information refers to an L1 motion vector and L1 It includes a picture number, and the L1 reference picture number may indicate a picture order count (POC) allocated to the L1 reference picture.

When the L0 motion vector and the L1 motion vector are the same and the L0 reference picture number and the L1 reference picture number are the same, the motion compensator is configured to: Based on the L0 motion information among the L0 motion information and the L1 motion information, Uni prediction can be performed.

When the L0 motion vector and the L1 motion vector are not identical to each other or the L0 reference picture number and the L1 reference picture number are not identical to each other, the motion compensator is configured to: Bi prediction can be performed.

Another embodiment of the present invention is an image decoding apparatus. The apparatus includes a motion predictor for deriving motion information of a current block as motion information including L0 motion information and L1 motion information, and a motion with respect to the current block based on at least one of the L0 motion information and the L1 motion information. A motion compensation unit generating a prediction block corresponding to the current block by performing compensation and a reconstructed block generation unit generating a reconstructed block corresponding to the current block based on the prediction block, wherein the motion compensator comprises: The motion compensation is performed based on the size of the current block.

When the size of the current block is smaller than a predetermined size, the motion compensator may perform uni prediction based on the L0 motion information among the L0 motion information and the L1 motion information.

The predetermined size may be 8x8.

Another embodiment of the present invention is an image decoding apparatus. The apparatus includes a motion prediction unit for deriving motion information from an already reconstructed picture, a motion compensation unit for generating a prediction block corresponding to the current block based on the derived motion information, and a prediction block for the current block based on the prediction block. a reconstructed block generating unit generating a corresponding reconstructed block, wherein the motion information includes a motion vector and a reference picture index, and the motion prediction unit sets the reference picture index to 0, and The motion vector is derived based on a col block corresponding to the current block in the reconstructed picture.

The encoding mode of the current block is a merge mode, and the motion compensator selects the temporal motion information as the motion information of the current block and performs motion compensation based on the selected motion information, so that the current block It is possible to generate a prediction block corresponding to .

According to the image encoding method according to the present invention, image encoding efficiency can be improved.

According to the image decoding method according to the present invention, image encoding efficiency can be improved.

According to the inter prediction method according to the present invention, image encoding efficiency can be improved.

According to the method for deriving motion information according to the present invention, image encoding efficiency can be improved.

According to the method for deriving temporal motion information according to the present invention, image encoding efficiency can be improved.

1 is a block diagram illustrating a configuration of an image encoding apparatus to which the present invention is applied according to an embodiment.
2 is a block diagram illustrating a configuration of an image decoding apparatus to which the present invention is applied according to an embodiment.
3 is a flowchart schematically illustrating an embodiment of an inter prediction method.
4 is a diagram schematically illustrating an embodiment of an inter prediction method when bidirectional prediction is applied.
5 is a diagram schematically illustrating an embodiment of motion information of an encoded image.
6 is a flowchart schematically illustrating an embodiment of a method for deriving temporal motion information of a current block according to the present invention.
7 is a diagram schematically illustrating an embodiment of a reconstructed neighboring block used for resetting L1 temporal motion information.
8 is a block diagram schematically illustrating an embodiment of an inter prediction apparatus capable of performing a process of deriving temporal motion information according to the embodiment of FIG. 6 .
9 is a flowchart schematically illustrating another embodiment of a method for deriving temporal motion information of a current block according to the present invention.
10 is a diagram schematically illustrating an embodiment of a second collocated block used for resetting L1 temporal motion information.
11 is a block diagram schematically illustrating an embodiment of an inter prediction apparatus capable of performing a process of deriving temporal motion information according to the embodiment of FIG. 9 .
12 is a diagram schematically illustrating an embodiment of a method for deriving motion information of a current block according to the present invention.
13 is a block diagram schematically illustrating an embodiment of an inter prediction apparatus capable of performing a process of deriving motion information according to the embodiment of FIG. 12 .
14 is a flowchart schematically illustrating another embodiment of a method for deriving temporal motion information of a current block according to the present invention.
15 is a block diagram schematically illustrating an embodiment of an inter prediction apparatus capable of performing a process of deriving temporal motion information according to the embodiment of FIG. 14 .

EMBODIMENT OF THE INVENTION Hereinafter, embodiment of this invention is concretely described 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.

When it is said that a component is “connected” or “connected” to another component, it may be directly connected or connected to the other component, but it is understood that other components may exist in between. it should be In addition, the content described as “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, second, etc. may be used to describe various elements, but the elements should not be limited by the terms. The above terms are used only for the purpose of distinguishing one component from another. For example, without departing from the scope of the present invention, a first component may be referred to as a second component, and similarly, a second component may also be referred to as a first component.

In addition, the components shown in the embodiment of the present invention are shown independently to represent different characteristic functions, and it does not mean that each component is composed of separate hardware or a single software component. That is, each component is listed as each component for convenience of description, and at least two components of each component are combined to form one component, or one component can be divided into a plurality of components to perform a function, and each Integrated embodiments and separate embodiments of components are also included in the scope of the present invention without departing from the essence of the present invention.

In addition, some of the components are not essential components for performing essential functions in the present invention, but may be optional components for merely improving performance. The present invention can be implemented by including only essential components to implement the essence of the present invention, except for 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 illustrating a configuration of an image encoding apparatus to which the present invention is applied according to an embodiment.

Referring to FIG. 1 , the image encoding apparatus 100 includes a motion prediction unit 111 , a motion compensation unit 112 , an intra prediction unit 120 , a switch 115 , a subtractor 125 , and a transform unit 130 . , a quantizer 140 , an entropy encoder 150 , an inverse quantizer 160 , an inverse transform unit 170 , an adder 175 , a filter unit 180 , and a 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 refers to intra prediction, and inter prediction refers to inter prediction. In the intra mode, the switch 115 may be switched to the intra mode, and in the inter mode, the switch 115 may be switched to the inter mode. The image encoding apparatus 100 may generate a prediction block for an input block of an input image, and then encode a residual between the input block and the prediction block.

In the intra mode, the intra prediction unit 120 may generate a prediction block by performing spatial prediction using pixel values of an already encoded block around the current block.

In 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 image stored in the reference picture buffer 190 during the motion prediction process. have. 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 the difference between the input block and the generated prediction block. The transform unit 130 may perform transform on the residual block to output a transform coefficient. In addition, the quantization unit 140 may quantize the input transform coefficient according to the quantization parameter to output a quantized coefficient.

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

When entropy encoding is applied, a small number of bits are allocated to a symbol having a high probability of occurrence and a large number of bits are allocated to a symbol having a low probability of occurrence to express the symbol, so that bits for symbols to be encoded The size of the column may be reduced. Accordingly, compression performance of image encoding may be improved through entropy encoding. The entropy encoder 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 image encoding apparatus according to the embodiment of FIG. 1 performs inter prediction encoding, that is, inter-frame prediction encoding, the currently encoded image needs to be decoded and stored to be used as a reference image. Accordingly, the quantized coefficients are inverse quantized by the inverse quantization unit 160 and inversely transformed by the inverse transform unit 170 . The inverse-quantized and inverse-transformed coefficients are added to the prediction block through the adder 175, and a reconstructed block is generated.

The reconstructed block passes through the 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 be referred to as an adaptive in-loop filter. The deblocking filter may remove block distortion and/or blocking artifacts generated at the boundary between blocks. The SAO may add an appropriate offset value to the pixel value to compensate for the coding error. ALF may perform filtering based on a comparison value between the reconstructed image and the original image, and may be performed only when high efficiency is applied. The reconstructed block passing through the filter unit 180 may be stored in the reference picture buffer 190 .

2 is a block diagram illustrating a configuration of an image decoding apparatus to which the present invention is applied 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 , and an adder. 255 , a filter unit 260 and a reference picture buffer 270 .

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

The entropy decoding unit 210 may entropy-decode the input bitstream according to a probability distribution to generate symbols including symbols in the form of quantized coefficients. 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 having a high occurrence probability and a large number of bits are allocated to a symbol having a low occurrence probability to represent the symbol, so that the size of the bit stream for each symbol is can be reduced. Accordingly, compression performance of image decoding may be improved through the entropy decoding method.

The quantized coefficient is inverse quantized by the inverse quantizer 220 and inverse transformed by 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 an already decoded block around the current block. In the inter mode, the motion compensator 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 may be 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 reconstructed 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 image encoding and decoding. When encoding and decoding an image, an encoding or decoding unit means a divided unit when encoding or decoding an image by dividing an image. Therefore, a coding unit (CU), a prediction unit (PU: prediction unit), and a transformation unit (TU) : Transform Unit), may be called a transform block (transform block), etc. One block may be further divided into smaller sub-blocks. In addition, in the present specification, “picture” may be used instead of “frame”, “field” and/or “slice” depending on the context, and such a distinction can be easily made by those of ordinary skill in the art. There will be. For example, a P picture, a B picture, and a forward B picture to be described later may be replaced with a P slice, a B slice, and a forward B slice, respectively, depending on context.

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

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

In the inter mode, the encoder and the decoder may derive motion information of the current block and then perform inter prediction and/or motion compensation based on the derived motion information. At this time, the encoder and the decoder obtain the motion information of the collocated block corresponding to the current block in the reconstructed neighboring block and/or the collocated picture that has already been reconstructed. By using it, encoding/decoding efficiency can be improved. Here, the reconstructed neighboring block is a block in the reconstructed current picture that has already been encoded and/or decoded, 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 the decoder may determine a predetermined relative position based on a block existing at the same spatial position as the current block in the collocated picture, and the determined relative position (existing at the same spatial position as the current block) The collocated block can be derived based on the position inside and/or outside the block). Here, as an example, the collocated picture may correspond to one of the reference pictures included in the reference picture list.

Meanwhile, the method of deriving motion information may vary depending on the prediction mode of the current block. Prediction modes applied for inter prediction may include Advanced Motion Vector Predictor (AMVP), merge, and the like.

For example, when Advanced Motion Vector Predictor (AMVP) is applied, the encoder and decoder may generate a predictive motion vector candidate list by using a motion vector of a reconstructed neighboring block and/or a motion vector of a collocated block. That is, the motion vector of the reconstructed neighboring block and/or the motion vector of the collocated block may be used as a prediction 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 the predicted motion vector of the current block from among the predicted motion vector candidates included in the predicted motion vector candidate list by using the predicted motion vector index.

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

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

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

In the aforementioned AMVP and merge modes, in order to derive the motion information of the current block, motion information of a reconstructed neighboring block and/or motion information of a collocated block may be used. Hereinafter, in embodiments to be described below, motion information derived from a reconstructed neighboring block is referred to as spatial motion information, and motion information derived based on a collocated block is referred to as temporal motion information. For example, a motion vector derived from a reconstructed neighboring block may be called a spatial motion vector, and a motion vector derived based on a collocated block may be called a temporal motion vector.

Referring back to FIG. 3 , the encoder and the 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 mean a motion-compensated block generated as a result of performing motion compensation on the current block. Also, a plurality of motion-compensated blocks may constitute one motion-compensated image. Accordingly, in the embodiments to be described below, the prediction block may be referred to as a 'motion compensated block' and/or a 'motion compensated image' depending on context, and this classification is easy for those of ordinary skill in the art. will be able to do

Meanwhile, a picture on which inter prediction is performed may include a P picture and a B picture. A P picture may mean a picture on which unidirectional prediction using one reference picture is performed, and a B picture may mean a picture on which forward, backward or bidirectional prediction using one or more, for example, two reference pictures can be performed. can do. For example, in the B picture, inter prediction may be performed using one forward reference picture (past picture) and one backward reference picture (future picture). Also, in the B picture, prediction may be performed using two forward reference pictures or prediction may be performed using two backward reference pictures.

Here, the reference pictures may be managed by a reference picture list. In the P picture, one reference picture is used, and the reference picture may be assigned to a reference picture list 0 (L0 or List0). In the B picture, two reference pictures are used, and the two reference pictures may be allocated to a reference picture list 0 and a reference picture list 1 (L1 or List1), respectively. Hereinafter, the L0 reference picture list may have the same meaning as the reference picture list 0, and the L1 reference picture list may have the same meaning as the reference picture list 1.

In general, the forward reference picture may be allocated to the reference picture list 0 and the backward reference picture may be allocated to the reference picture list 1 . However, the method of allocating the reference picture is not limited thereto, and the forward reference picture may be allocated to the reference picture list 1 and the backward reference picture may be allocated to the reference picture list 0. Hereinafter, the reference picture allocated to the reference picture list 0 is referred to as an L0 reference picture, and the reference picture allocated to the reference picture list 1 is referred to as an L1 reference picture.

In general, reference pictures may be allocated to the reference picture list in descending order according to the reference picture number. Here, the reference picture number may mean a number allocated to each reference picture in a picture order count (POC) order, and the POC order may mean a display order and/or a temporal order of pictures. For example, two reference pictures having the same reference picture number may correspond to the same reference picture. The reference pictures allocated to the reference picture list may be rearranged by a reference picture list reordering (RPLR) or a memory management control operation (MMCO) command.

As described above, unidirectional prediction using one L0 reference picture may be performed in the P picture, and one L0 reference picture and one L1 reference picture in the B picture, that is, forward, backward, or bidirectional using two reference pictures. Prediction may be performed. Prediction using one reference picture may be called uni-prediction, and prediction using two reference pictures including the L0 reference picture and the L1 reference picture may be called bi-prediction.

Bi-prediction can be used as a concept including all forward prediction, backward prediction, and bi-prediction, but in embodiments to be described below, for convenience, prediction using two reference pictures (L0 reference picture and L1 reference picture) is called bi-directional prediction. That is, in the embodiments to be described later, bi-prediction may mean bi-prediction, and it may be understood as a concept including forward, backward, and bi-prediction using two reference pictures (L0 reference picture and L1 reference picture). . In addition, forward prediction or backward prediction may be performed even when bi-prediction is performed, but in embodiments to be described below, prediction using only one reference picture is referred to as uni-directional prediction for convenience. That is, in embodiments to be described later, uni-prediction may mean uni-prediction, and should be understood as a concept including only prediction using one reference picture. Also, hereinafter, information indicating whether uni-prediction (uni-prediction) or bi-prediction (bi-prediction) is applied to a block on which prediction is performed is referred to as prediction direction information.

4 is a diagram schematically illustrating an embodiment of an inter prediction method when bidirectional prediction is applied.

As described above, the encoder and the decoder may perform bidirectional prediction as well as unidirectional prediction during inter prediction. When bidirectional prediction is applied, each block on which prediction is performed may have two reference pictures (L0 reference picture and L1 reference picture). Also, at this time, each block on which bidirectional prediction is performed may have two pieces of motion information. Here, the motion information may include a reference picture number and a motion vector.

When bidirectional prediction is performed, the encoder and the decoder may select one reference picture from the reference picture list 0 and the reference picture list 1, respectively, and use it for prediction. That is, two reference pictures including the L0 reference picture and the L1 reference picture may be used for bi-directional prediction. Hereinafter, motion information corresponding to the L0 reference picture will be referred to as L0 motion information, and motion information corresponding to the L1 reference picture will be referred to as L1 motion information. Also, motion compensation using L0 motion information is called L0 motion compensation, and motion compensation using L1 motion information is called L1 motion compensation.

Referring to FIG. 4 , the encoder and the decoder may generate the L0 motion-compensated block by performing the L0 motion compensation 410 on the current block using the L0 motion information and the L0 reference picture list. Also, the encoder and the decoder may generate the L1 motion-compensated block by performing the L1 motion compensation 420 using the L1 motion information and the L1 reference picture list. In this case, the L0 motion compensation 410 and L1 motion compensation 420 processes may be performed independently of each other.

The encoder and the decoder may perform a weighted average 430 on the L0 motion-compensated block and the L1 motion-compensated block to finally generate one motion-compensated block. For example, the weighted average 430 may be performed in units of pixels in the L0 motion-compensated block and the L1 motion-compensated block. In this case, one finally generated motion-compensated block may correspond to a prediction block of the current block.

Hereinafter, motion compensation applied during bidirectional prediction is referred to as bidirectional motion compensation. Correspondingly, motion compensation applied in unidirectional prediction may be referred to as unidirectional motion compensation.

5 is a diagram schematically illustrating an embodiment of motion information of an encoded image. 5 illustrates a plurality of blocks constituting an encoded image and motion information of each of the plurality of blocks.

The encoded image in FIG. 5 corresponds to BasketballDrill. Here, BasketballDrill represents the name of a test sequence used in an image encoding/decoding experiment. The size of the encoded image is 832x480, and the picture order count (POC) is 2. In addition, a quantization parameter (QP) value applied to the image of FIG. 5 is 32.

The encoder and the decoder may use the forward B picture in order to increase inter prediction efficiency in a low-latency application environment. Here, the forward B picture may mean a B picture on which only forward prediction is performed. When the forward B picture is used, each block on which prediction is performed may have two pieces of motion information (L0 motion information and L1 motion information). In a forward B picture, in general, the L0 reference picture list and the L1 reference picture list may be set identically. Hereinafter, in the present specification, when the forward B picture is used, it is assumed that the L0 reference picture list and the L1 reference picture list are the same.

* The decoder may directly determine whether the current picture is a forward B picture based on the L0 reference picture list and the L1 reference picture list, but based on the information transmitted from the encoder, it is possible to determine whether the current picture is a forward B picture. may be For example, the encoder may encode a flag indicating whether the L0 reference picture list and the L1 reference picture list are the same and transmit it to the decoder. In this case, after receiving and decoding the flag, the decoder may determine whether the current picture is a forward B picture based on the decoded flag. As another example, the encoder may transmit the NAL unit type value or the slice type value corresponding to the forward B picture to the decoder, and the decoder may receive the value and determine whether it is a forward B picture based on the received value.

The image shown in FIG. 5 is an image encoded using a forward B picture. Accordingly, each block in the coded image may have up to two pieces of motion information. Here, the motion information may include a reference picture number, a motion vector, and the like. Referring to FIG. 5 , a block having the same L0 motion information (eg, reference picture number, motion vector) and L1 motion information (eg, reference picture number, motion vector) among blocks having two motion information is the same. There may be many.

A block having the same L0 motion information (eg, reference picture number, motion vector) and L1 motion information (eg, reference picture number, motion vector) in the forward B picture may occur due to the temporal motion information derivation method. have. As described above, the temporal motion information may be derived from motion information of the collocated block corresponding to the current block in the already reconstructed collocated picture. For example, when deriving the L0 temporal motion information of the current block, the encoder and the decoder may use the L0 motion information of the collocated block corresponding to the current block in the collocated picture. However, when the L0 motion information does not exist in the collocated block, the encoder and the decoder may use the L1 motion information of the collocated block as the L0 temporal motion information of the current block. Conversely, when deriving the L1 temporal motion information of the current block, the encoder and the decoder may use the L1 motion information of the collocated block corresponding to the current block in the collocated picture. However, when the L1 motion information does not exist in the collocated block, the encoder and the decoder may use the L0 motion information of the collocated block as the L1 temporal motion information of the current block. As a result of performing the above-described process, a phenomenon may occur in which the L0 motion information and the L1 motion information of the current block become the same. Accordingly, in the present specification, when L0 motion information (eg, reference picture number and motion vector) and L1 motion information (eg, reference picture number and motion vector) are the same, and/or L0 temporal motion information (eg, For example, when the reference picture number and the motion vector) and the L1 temporal motion information (eg, the reference picture number and the motion vector) are the same, the motion information of the collocated block corresponding to the current block in the already reconstructed collocated picture is L0. It may include a case of having only one of the motion information and the L1 motion information.

Also, when a block having the same L0 motion information and L1 motion information is generated, the block may affect a block to be encoded later. For example, when merge is applied, motion information (L0 motion information and L1 motion information) of the reconstructed neighboring block and/or collocated block may be used as the motion information of the current block as it is. Accordingly, when blocks having the same L0 motion information and L1 motion information are generated, more blocks having the same L0 motion information and L1 motion information may be generated.

When motion compensation is performed on a block having the same L0 motion information and L1 motion information, the same process may be repeatedly performed twice in one block. Since this is very inefficient in terms of encoding, an inter prediction method and/or motion compensation method capable of reducing computational complexity and improving encoding efficiency by solving the above-described problems may be provided. For example, when L0 motion information (eg, reference picture number and motion vector) and L1 motion information (eg, reference picture number and motion vector) are the same, the encoder and the decoder perform the motion compensation process only once. The computational complexity can be reduced. As another example, when the L0 motion information (eg, reference picture number and motion vector) and L1 motion information (eg, reference picture number and motion vector) are the same, the encoder and the decoder perform the L0 motion information and/or L1 motion information. It is also possible to increase the encoding efficiency by resetting the information.

6 is a flowchart schematically illustrating an embodiment of a method for deriving temporal motion information of a current block according to the present invention.

Embodiments to be described below will be described based on temporal motion information, but the present invention is not limited thereto. For example, the methods according to the embodiment of FIG. 6 include not only temporal motion information in the merge mode and/or AMVP mode, but also motion information of a current block derived based on the merge candidate list in the merge mode and/or in the AMVP mode. The same or similar method may be applied to motion information of the current block derived based on the prediction motion vector candidate list.

As described above, the temporal motion information may be derived based on the motion information of the collocated block corresponding to the current block in the already reconstructed collocated picture. Here, the collocated picture may correspond to, for example, one picture among reference pictures included in the reference picture list. The encoder and the decoder may determine a predetermined relative position based on a block existing at the same spatial position as the current block in the collocated picture, and the determined relative position (eg, spatially identical to the current block). The collocated block may be derived based on the position inside and/or outside the block existing in the position). The temporal motion information derived based on the collocated block may include prediction direction information, L0 reference picture number, L1 reference picture number, L0 motion vector, L1 motion vector, and the like.

Referring to FIG. 6 , the encoder and the decoder determine whether L0 temporal motion information and L1 temporal motion information are the same in the temporal motion information derived based on the collocated block, that is, the L0 reference picture number and the L1 reference picture number are the same and the L0 motion information is the same. It may be determined whether the vector and the L1 motion vector are the same (S610).

When the L0 temporal motion information and the L1 temporal motion information are not the same, that is, the L0 reference picture number and the L1 reference picture number are not the same and/or the L0 motion vector and the L1 motion vector are not the same, the encoder and the decoder The temporal motion information derived based on the block may be used as the temporal motion information for the current block as it is. When AMVP is applied, the temporal motion information of the current block may be determined or registered as a prediction motion vector candidate for the current block. Also, when the merge is applied, the temporal motion information of the current block may be determined or registered as a merge candidate for the current block.

When the L0 temporal motion information and the L1 temporal motion information are the same, the encoder and the decoder may determine whether motion information of a reconstructed neighboring block exists (S620). For example, at this time, the encoder and the decoder may determine whether a block having motion information exists among reconstructed neighboring blocks at a predetermined position and/or a position selected by a predetermined method. Here, the reconstructed neighboring block is a block in the reconstructed current picture that has already been encoded and/or decoded, and may include a block adjacent to the current block and/or a block located at an outer corner of the current block.

When motion information of a reconstructed neighboring block does not exist (eg, when a block having motion information does not exist among reconstructed neighboring blocks at a predetermined position and/or a position selected by a predetermined method), the encoder and the decoder The group may reset the prediction direction information of the current block to unidirectional prediction (S630). Also, at this time, the encoder and the decoder may use only the L0 temporal motion information as the temporal motion information of the current block.

When motion information of a reconstructed neighboring block exists (for example, when a block having motion information exists among reconstructed neighboring blocks at a predetermined position and/or a position selected by a predetermined method), the encoder and the decoder The motion information of the reconstructed neighboring block may be used as the L1 temporal motion information of the current block (S640). That is, at this time, the encoder and the decoder may reset the L1 temporal motion information of the current block to the motion information of the reconstructed neighboring block. A specific embodiment of the reconstructed neighboring block used for setting the L1 temporal motion information will be described later.

In the above-described embodiment, the method for deriving temporal motion information of the current block is described based on a flowchart as a series of steps, but one or more steps in the flowchart may be deleted. For example, in the above-described embodiment, steps S620 and S640 may be omitted. In this case, when the L0 temporal motion information and the L1 temporal motion information are the same, the encoder and the decoder may reset the prediction direction information of the current block to unidirectional prediction ( S630 ). Also, at this time, the encoder and the decoder may use only the L0 temporal motion information as the temporal motion information of the current block.

Meanwhile, in the above-described embodiment, it is determined whether to perform the processes S620 to S640 based on the sameness of the L0 temporal motion information and the L1 temporal motion information, but the encoder and the decoder determine whether to perform the processes S620 to S640 based on other conditions. may be

As an embodiment, the encoder and the decoder may determine whether to perform steps S620 to S640 based on the sameness of the L0 reference picture number and the L1 reference picture number in the temporal motion information derived based on the collocated block. For example, when the L0 reference picture number and the L1 reference picture number are the same, the encoder and the decoder may perform processes S620 to S640.

As another embodiment, the encoder and the decoder may determine whether to perform steps S620 to S640 based on the prediction direction of the collocated block. As described above, the prediction direction information may refer to information indicating whether unidirectional prediction or bidirectional prediction is applied to a block on which prediction is performed. Accordingly, the prediction direction may correspond to unidirectional prediction or bidirectional prediction. For example, when the motion information (prediction direction) of the collocated block is unidirectional prediction rather than bidirectional prediction, the encoder and the decoder may perform processes S620 to S640. This is because, in the case where the prediction direction of the collocated block is unidirectional prediction, as a result, in the temporal motion information derived based on the collocated block, the L0 temporal motion information and the L1 temporal motion information may be the same.

As another embodiment, the encoder and the decoder may determine whether to perform steps S620 to S640 based on information on whether motion information exists in the collocated block. For example, the encoder and the decoder may perform steps S620 to S640 when motion information does not exist in the collocated block. In this case, in step S640 described above, the L0 temporal motion information, not the L1 temporal motion information, may be reset. That is, the encoder and the decoder may set the L0 temporal motion information as the motion information of the reconstructed neighboring block, and may perform unidirectional prediction instead of bidirectional prediction on the current block. Also, when motion information does not exist in the collocated block, the encoder and the decoder may reset both the L0 temporal motion information and the L1 temporal motion information in step S640 described above. That is, the encoder and the decoder may set the L0 temporal motion information and the L1 temporal motion information as motion information of the reconstructed neighboring block, and may perform bidirectional prediction on the current block.

As another embodiment, the encoder and the decoder may determine whether to perform steps S620 to S640 based on the size of the current block. For example, the encoder and the decoder may determine whether the size of the current block is smaller than a predetermined size. Here, the current block may be a CU, a PU, and/or a TU, and the predetermined size may be, for example, one of 8x8, 16x16, or 32x32. In this case, when the size of the current block is smaller than a predetermined size, the encoder and the decoder may perform processes S620 to S640.

As another embodiment, the encoder and the decoder may perform steps S620 to S640 when the L0 motion vector and/or the L1 motion vector in the motion information of the collocated block corresponds to a zero vector (0,0). In this case, in step S640 described above, the encoder and the decoder may reset the motion vector(s) corresponding to the zero vector (0,0). As an example, the motion vector(s) corresponding to the zero vector (0,0) may be set as a motion vector of a reconstructed neighboring block, and as another example, the motion vector(s) corresponding to the zero vector (0,0) ( ) may be set as a motion vector of a block located in the vicinity of the collocated block. As another embodiment, the encoder and the decoder may perform steps S620 to S640 when the L0 motion vector and/or the L1 motion vector in the motion information of the collocated block do not correspond to the zero vector (0,0). In this case, in step S640 described above, the encoder and the decoder may reset motion vector(s) that do not correspond to the zero vector (0,0), and the motion vector(s) that do not correspond to the zero vector (0,0) ) may be reset to the motion vector of the reconstructed neighboring block.

The conditions for determining whether to perform the processes S620 to S640 are not limited to the above-described embodiments, and various conditions may be applied according to conditions and/or needs.

On the other hand, when AMVP is applied, the encoder and/or decoder may use the candidate with the smallest difference from the motion vector of the current block among the motion vector candidates in the prediction motion vector candidate list as the predicted motion vector of the current block, and the merge is applied. In this case, the encoder and/or decoder may use a merge candidate capable of providing optimal encoding efficiency among merge candidates in the merge candidate list as motion information for the current block. Since the prediction motion vector candidate list and the merge candidate list may include temporal motion information and/or spatial motion information, respectively, as a result, motion information (eg, reference picture number, motion information, etc.) of the current block is restored around It can be derived based on the motion information of the block. Therefore, in the above-described L1 temporal motion information resetting step S640 , the encoder and the decoder may reset the L1 temporal motion information based on the predicted motion vector candidate list of the current block and/or the merge candidate list of the current block. In this case, the step ( S630 ) of determining whether the above-described reconstructed neighboring blocks exist may be omitted. Hereinafter, in an embodiment to be described later, the motion information candidate list may be understood as a concept including a predictive motion vector candidate list and a merge candidate list.

As described above, the encoder and the decoder may reset the L1 temporal motion information of the current block based on the motion information candidate list of the current block. That is, the encoder and the decoder may use one of the motion information candidates included in the motion information candidate list of the current block as the L1 temporal motion information. Motion information used as L1 temporal motion information in the motion information candidate list of the current block may be determined in various ways.

As an embodiment, the encoder and the decoder may sequentially search for motion information candidates in the motion information candidate list from a first motion information candidate to a last motion information candidate. In this case, the encoder and the decoder may use the first motion information candidate having the same reference picture number as the L1 reference picture number of the L1 temporal motion information derived based on the collocated block as the L1 temporal motion information of the current block. In this case, the encoder and the decoder may use the L0 motion information of the first motion information candidate having the same reference picture number as the L1 reference picture number, and the first motion information candidate having the same reference picture number as the L1 reference picture number. It is also possible to use L1 motion information of . Meanwhile, motion information having the same reference picture number as the L1 reference picture number of the L1 temporal motion information derived based on the collocated block may not exist in the motion information candidate list. In this case, the encoder and the decoder may add motion information having the same reference picture number as the L1 reference picture number among the motion information of the reconstructed neighboring blocks to the motion information candidate list. In this case, the encoder and the decoder may reset the L1 temporal motion information of the current block based on the motion information candidate list to which the motion information of the reconstructed neighboring block is added.

As another embodiment, the encoder and the decoder may use the motion vector of the first motion information candidate having the available motion vector in the motion information candidate list as the L1 temporal motion information of the current block. In this case, the encoder and the decoder may use the L0 motion information of the first motion information candidate or the L1 motion information of the first motion information candidate.

As another embodiment, the encoder and the decoder may use the first motion information candidate available in the motion information candidate list as the L1 temporal motion information of the current block. In this case, the L1 reference picture number of the L1 temporal motion information is changed or set to the reference picture number of the first motion information candidate, and the L1 motion vector of the L1 temporal motion information is changed to the motion vector of the first motion information candidate, or can be set. In this case, the encoder and the decoder may use the L0 motion information of the first motion information candidate or the L1 motion information of the first motion information candidate.

A method of determining a motion information candidate used as L1 temporal motion information in the motion information candidate list of the current block is not limited to the above-described embodiment, and various conditions may be applied according to conditions and/or needs.

In the above-described embodiments, there may be no motion information available in the motion information candidate list of the current block. In this case, as an example, the encoder and the decoder may use the motion information of the reconstructed neighboring block as the L1 temporal motion information of the current block, as in step S640 described above. As another example, in this case, the encoder and the decoder may add the motion information of the reconstructed neighboring block to the motion information candidate list, and may add a zero vector (0,0) to the motion information candidate list. In this case, the encoder and the decoder may reset the L1 temporal motion information of the current block based on the motion information candidate list to which the motion information of the reconstructed neighboring block is added or the motion information candidate list to which the zero vector (0,0) is added. have.

In the above-described embodiments, since the L0 temporal motion information and the L1 temporal motion information are temporally derived motion information, there is a high possibility that they correspond to motion information due to the movement of an object. Accordingly, when the encoder and the decoder search for motion information to be used as L1 temporal motion information of the current block from the reconstructed neighboring block and/or motion information candidate list, motion information having a zero vector (0,0) is not selected and zero It is also possible to select motion information that does not have a vector (0,0). This is because a block having motion information corresponding to a zero vector (0, 0) is highly likely to correspond to a background rather than an object.

Meanwhile, the reset L1 temporal motion information may be the same as the L0 temporal motion information of the current block. Accordingly, the encoder and the decoder may select motion information that is not identical to the L0 temporal motion information when searching for motion information to be used as the L1 temporal motion information of the current block from the reconstructed neighboring block and/or motion information candidate list. For example, as in S640 described above, when the L1 temporal motion information of the current block is derived based on the reconstructed neighboring block, the encoder and the decoder may select a block having motion information different from the L0 temporal motion information of the current block. It can be determined as a block used for deriving L1 temporal motion information. In this case, the encoder and the decoder may select only motion information whose difference from the L0 temporal motion information of the current block is less than or equal to a predetermined threshold as motion information to be used as the L1 temporal motion information of the current block. Here, the predetermined threshold value may be determined based on mode information of the current block, motion information of the current block, mode information of the neighboring block, and/or motion information of the neighboring block, and may be determined in various ways.

Also, in the above-described embodiments, the motion information of the reconstructed neighboring block and the motion information selected from the motion information candidate list may include both L0 motion information and L1 motion information, respectively. In this case, the encoder and the decoder may determine one of the L0 motion information and the L1 motion information as motion information to be used as the L1 temporal motion information of the current block. For example, the encoder and the decoder may use the L0 motion information from the motion information of the reconstructed neighboring block and/or the motion information selected from the motion information candidate list as the L1 temporal motion information of the current block. In this case, when the L0 motion information does not exist in the motion information of the reconstructed neighboring block and/or the motion information selected from the motion information candidate list, the encoder and the decoder may use the L1 motion information as the L1 temporal motion information of the current block. . As another example, the encoder and the decoder may use the L1 motion information from the motion information of the reconstructed neighboring block and/or the motion information selected from the motion information candidate list as the L1 temporal motion information of the current block. In this case, when the L1 motion information does not exist in the motion information of the reconstructed neighboring block and/or the motion information selected from the motion information candidate list, the encoder and the decoder may use the L0 motion information as the L1 temporal motion information of the current block. .

On the other hand, the encoder and the decoder do not use the motion information and/or the motion information candidate list of the reconstructed neighboring block to reset the L1 temporal motion information (eg, the L1 motion vector) of the current block in step S640 described above. may be In this case, the encoder and the decoder may reset L1 temporal motion information (eg, L1 motion vector) of the current block based on L0 temporal motion information (eg, L0 motion vector) of the current block. Hereinafter, related embodiments will be described, which will be described based on a motion vector.

In an embodiment, the encoder and the decoder transmit motion information indicating a relative position moved based on a predetermined distance and/or direction from a position indicated by the L0 temporal motion information (L0 motion vector) of the current block to the L1 of the current block. It can be used as temporal motion information (L1 motion vector). As an example, the encoder and the decoder have a 1/4 pixel size (eg, (-1,0), (1,0), ( 0,-1), (0,1), (-1,-1), (-1,1), (1,-1), (1,1), etc. where 1/4 pixel unit is 1 ) may be used as the L1 temporal motion information of the current block. As another example, the encoder and the decoder have a 1/2 pixel size (eg, (-2,0), (2,0), ( 0,-2), (0,2), (-2,-2), (-2,2), (2,-2), (2,2), etc. where 1/4 pixel unit is 1 ) may be used as the L1 temporal motion information of the current block. As another example, the encoder and the decoder have integer pixel sizes (eg, (-4,0), (4,0), (0) in the vertical and/or horizontal direction at the position indicated by the L0 temporal motion information of the current block. ,-4), (0,4), (-4,-4), (-4,4), (4,-4), (4,4), etc. where 1/4 pixel unit is 1 The motion information indicating the position moved by the corresponding amount) may be used as the L1 temporal motion information of the current block. On the other hand, since the motion vector includes a vertical component and a horizontal component, the above-described methods (moving by 1/4 pixel size, moving by 1/2 pixel size, moving by integer pixel size) include a horizontal component and a vertical component. Each component may be independently applied. In this case, the moving distance in the horizontal direction and the moving distance in the vertical direction may be different from each other.

In another embodiment, the encoder and the decoder change the value of the L0 temporal motion information (L0 motion vector) of the current block to a value of another pixel unit, and then use the changed value as the L1 temporal motion information (L1 motion vector) of the current block. can As an example, when the value of the L0 temporal motion information of the current block is a value in units of 1/4 pixels, the encoder and the decoder can change the value of the L0 temporal motion information to a value in units of 1/2 pixels based on a shift operation, etc. and the changed value may be used as L1 temporal motion information of the current block. As another example, if the value of the L0 temporal motion information of the current block is a value of 1/2 pixel unit, the encoder and the decoder may change the value of the L0 temporal motion information to a value of integer pixel units based on a shift operation, etc., The changed value may be used as L1 temporal motion information of the current block.

The method of resetting the L1 temporal motion information (eg, L1 motion vector) of the current block based on the L0 temporal motion information (eg, L0 motion vector) of the current block is not limited to the above-described embodiment, and the implementation and/or may be applied in various forms as needed.

Meanwhile, in the above-described embodiment, the temporal motion information input to step S610 before resetting the temporal motion information may include not only a motion vector but also a reference picture index. Here, the L0 motion vector and the L1 motion vector may be motion vectors temporally derived based on the collocated block as described above, and the L0 reference picture index and the L1 reference picture index are spatially derived reference pictures from the reconstructed neighboring blocks. It can be an index. In this case, the L0 reference picture index and the L1 reference picture index may be set to the smallest non-negative value among the reference picture indexes of the reconstructed neighboring blocks. Meanwhile, as another example, the L0 input reference picture index and the L1 input reference picture index may be set to 0 regardless of the motion information of the reconstructed neighboring blocks.

When the L1 motion vector of the L1 temporal motion information is reset based on the reconstructed neighboring block, the resetting process may also be performed on the L1 reference picture index of the L1 temporal motion information.

Hereinafter, only for the embodiments of FIGS. 6 to 8, which will be described later, for convenience of explanation, the temporal motion information input to step S610 before resetting the temporal motion information is called input temporal motion information (L0 input temporal motion information, L1 input temporal motion information). . In addition, the motion vector included in the input temporal motion information is an input motion vector (L0 input motion vector, L1 input motion vector), and the reference picture index included in the input temporal motion information is an input reference picture index (L0 input reference picture index, L1 input). A reference picture index) and a reference picture number included in the input temporal motion information are referred to as an input reference picture number (L0 input reference picture number, L1 input reference picture number).

As described above, when the L1 input motion vector is reset, the encoder and the decoder may also perform a reset process on the L1 input reference picture index. Hereinafter, embodiments of the reset process for the L1 input reference picture index will be described.

As an embodiment, the encoder and the decoder may reset the L1 input reference picture index based on the reconstructed neighboring blocks. As described above, the encoder and the decoder may use the reconstructed motion information of the neighboring block as the L1 temporal motion information of the current block. In this case, the encoder and the decoder may derive the final L1 reference picture index by resetting the L1 input reference picture index to the reference picture index of the reconstructed neighboring block. In another embodiment, the encoder and the decoder may derive the final L1 reference picture index by resetting the L1 input reference picture index value to a predetermined fixed reference picture index value.

In another embodiment, L0 input temporal motion information (eg, L0 input motion vector, L0 input reference picture index, etc.) and L1 input temporal motion information (eg, L1 input motion vector, L1 input reference picture index, etc.) ) is the same, the encoder and the decoder may reset both the L0 input reference picture index and the L1 input reference picture index to a value of 0 to use as final temporal motion information. This is because, when the L0 input temporal motion information and the L1 input temporal motion information are the same, there is a high probability that both the L0 reference picture index and the L1 reference picture index are 0.

As another embodiment, the encoder and the decoder may reset the L1 input reference picture index to the same value as the L0 input reference picture index. On the other hand, the encoder and the decoder may reset the L1 input reference picture index value to a reference picture index value that is not the same as the L0 input reference picture index. As an example, a reference picture index that is not the same as the L0 input reference picture index may exist among the reference picture indexes of the reconstructed neighboring blocks. In this case, for example, the encoder and the decoder may use the most frequently used reference picture index among reference picture indexes that are not the same as the L0 input reference picture index to reset the L1 input reference picture index. As another example, in this case, the encoder and the decoder may use the reference picture index having the smallest non-negative value among the reference picture indices that are not the same as the L0 input reference picture index to reset the L1 input reference picture index.

Meanwhile, as described above, the encoder and the decoder may derive the final L1 temporal motion vector by resetting the L1 input motion vector value in the L1 input temporal motion information to the same value as the motion vector of the reconstructed neighboring block. In this case, the motion vector of the reconstructed neighboring block may be scaled and used according to the L1 input reference picture index and/or the reset L1 reference picture index. In the L1 input temporal motion information, the L1 input reference picture index may be used as the final L1 temporal motion information as it is without a resetting process, or may be used as the final L1 temporal motion information through a resetting process as in the above-described embodiment. In this case, the reference picture corresponding to the motion vector of the reconstructed neighboring block and the reference picture indicated by the final L1 reference picture index may be different from each other. In this case, the encoder and the decoder may perform scaling on the motion vector of the reconstructed neighboring block and use the scaled motion vector as the final L1 temporal motion information of the current block.

The above-described embodiments may be combined and applied in various ways according to a motion vector resetting process and a reference picture index resetting process (eg, RefIdxLX, X=0,1). Hereinafter, it is assumed that the L1 input motion vector is reset based on the reconstructed neighboring block in the embodiments to be described below. That is, it is assumed that the encoder and the decoder reset the L1 input motion vector to one of the motion vectors of the reconstructed neighboring blocks.

In an embodiment, the encoder and the decoder may use only a motion vector other than a zero vector (0,0) among motion vectors of reconstructed neighboring blocks to reset the L1 input motion vector. In this case, for example, the L1 input reference picture index may be used as the final L1 temporal motion information as it is without a reset process. As another example, the L1 input reference picture index value may be reset to the reference picture index value of the reconstructed neighboring block, and the reset reference picture index may be used as the final L1 temporal motion information. As another example, the L1 input reference picture index value may be reset to a predetermined fixed reference picture index value. Specific examples of each of these have been described above, and thus will be omitted here.

In another embodiment, the encoder and the decoder may scale and use the motion vector of the reconstructed neighboring block to reset the L1 input motion vector. In this case, the scaled motion vector may be used as the final L1 temporal motion information. In this case, for example, the L1 input reference picture index may be used as the final L1 temporal motion information as it is without a reset process. As another example, the L1 input reference picture index value may be reset to the reference picture index value of the reconstructed neighboring block, and the reset reference picture index may be used as the final L1 temporal motion information. As another example, the L1 input reference picture index value may be reset to a predetermined fixed reference picture index value. Specific examples of each of these have been described above, and thus will be omitted here.

In another embodiment, in order to reset the L1 input motion vector, the encoder and the decoder only use motion vectors whose difference from the L0 temporal motion information (L0 motion vector) of the current block is less than or equal to a predetermined threshold among motion vectors of the reconstructed neighboring blocks. Can be used. In this case, the encoder and the decoder may use only a motion vector that is not identical to the L0 temporal motion information (L0 motion vector) of the current block among the motion vectors of the reconstructed neighboring blocks. In this case, for example, the L1 input reference picture index may be used as the final L1 temporal motion information as it is without a reset process. As another example, the L1 input reference picture index value may be reset to the reference picture index value of the reconstructed neighboring block, and the reset reference picture index may be used as the final L1 temporal motion information. As another example, the L1 input reference picture index value may be reset to a predetermined fixed reference picture index value. Specific examples of each of these have been described above, and thus will be omitted here.

The combination of the embodiments of the motion vector resetting process and the reference picture index resetting process is not limited to the above-described embodiment, and various types of combinations may be provided as well as the above-described embodiments according to implementation and/or necessity. have.

Meanwhile, the L1 temporal motion information reset as described above may be the same as the L0 temporal motion information of the current block. Accordingly, when the reset L1 temporal motion information is the same as the L0 temporal motion information of the current block, the encoder and the decoder may reset the prediction direction information of the current block to unidirectional prediction. In this case, the encoder and the decoder may use only the L0 temporal motion information as the temporal motion information of the current block. This method can be applied to the present invention in combination with the above-described embodiments.

7 is a diagram schematically illustrating an embodiment of a reconstructed neighboring block used for resetting L1 temporal motion information. In FIG. 7, block 710 represents the current block (X).

As described above, the reconstructed neighboring block is a block in the reconstructed current picture that has already been encoded and/or decoded, and may include a block adjacent to the current block 710 and/or a block located at an outer corner of the current block. In the embodiment of FIG. 7 , the block located at the lower left corner outside the current block 710 is called a lower left corner block (A), and the block located at the upper left corner outside the current block 710 is the upper left corner block (E) ), and the block located in the upper right corner outside the current block 710 is referred to as the upper right corner block (C). Also, among the blocks adjacent to the left of the current block 710, the uppermost block is the upper left block (F), the lowermost block among the blocks adjacent to the left of the current block 710 is the left lowermost block (A), the current block ( 710) Among the blocks adjacent to the top, the leftmost block is called the top leftmost block (G), and the rightmost block among the blocks adjacent to the top of the current block 710 is called the top rightmost block (B).

As described above in the embodiment of FIG. 6 , the encoder and the decoder may use the motion information of the neighboring block reconstructed according to a predetermined condition as the L1 temporal motion information of the current block. That is, the encoder and the decoder may reset the L1 temporal motion information value of the current block to the motion information value of the reconstructed neighboring block. In this case, the motion information used as the L1 temporal motion information of the current block may be derived in various ways.

As an embodiment, the encoder and the decoder may derive motion information used as the L1 temporal motion information from one block existing at a predetermined position among the reconstructed neighboring blocks. At this time, the block at the predetermined position is the lower left block (A), the upper right block (B), the upper right corner block (C), the lower left corner block (D), the upper left corner block (E), the left It may correspond to one of the uppermost block (F) and the upper leftmost block (G).

In another embodiment, the encoder and the decoder may scan a plurality of blocks existing at a predetermined position among the reconstructed neighboring blocks in a predetermined order. In this case, the encoder and the decoder may use the motion information of the first block in which the motion information exists in the scan order as the L1 temporal motion information of the current block. A scan target block and a scan order may be determined in various forms. As an example, the encoder and the decoder may scan the neighboring blocks in the order of the left uppermost block (F), the upper leftmost block (B), and the upper rightmost block (B). As another example, the encoder and decoder block neighboring blocks in the order of the upper left block (F), the upper leftmost block (G), the upper right corner block (C), the lower left corner block (D), and the upper left corner block (E). You can also scan. As another example, the encoder and the decoder may scan the neighboring blocks in the order of the lower left block (A), the upper right block (B), and the upper left corner block (E). As another example, the encoder and decoder are adjacent blocks in the order of the lower left block (A), the upper right block (B), the upper right corner block (C), the lower left corner block (D), and the upper left corner block (E) can also be scanned. As another example, the encoder and decoder are the lower left block (A), the upper right block (B), the upper right corner block (C), the lower left corner block (D), the upper left corner block (E), the upper left block ( F) and the upper leftmost block (G) in the order of the neighboring blocks may be scanned.

As another embodiment, the encoder and the decoder may derive an intermediate value for motion information of a plurality of blocks existing at a predetermined position among the reconstructed neighboring blocks, and use the derived intermediate value as the L1 temporal motion information of the current block. can be used as For example, the plurality of blocks may be a left uppermost block (F), an upper leftmost block (G), an upper right corner block (C), a lower left corner block (D), and an upper left corner block (E). .

The method of deriving the L1 temporal motion information of the current block from the reconstructed neighboring block is not limited to the above-described embodiment, and the L1 temporal motion information of the current block may be derived in various ways according to implementation and/or necessity.

Hereinafter, an embodiment of a process of deriving temporal motion information according to the embodiments of FIGS. 6 and 7 from the decoder's point of view will be described.

The input for the temporal motion information derivation process may include the position (xP, yP) of the upper-left sample of the current block, an input motion vector (mvLXCol), and an input reference picture number (RefPicOrder(currPic, refIdxLX, LX)). have. Here, currPic may mean a current picture. The output of the process may be a motion vector (mvLXCol) used as the final temporal motion information and information indicating the presence or absence of the final temporal motion information (availableFlagLXCol). Here, X may have a value of 0 or 1. For example, when X is 0, mvLXCol, refIdxLX, and LX may represent mvL0Col, refIdxL0, and L0, which may mean variables related to L0 temporal motion information. Also, mvLX may indicate a motion vector, mvLX[0] may indicate an x-component motion vector, and mvLX[1] may indicate a y-component motion vector. refIdxLX may indicate an LX reference picture index indicating a reference picture in an LX reference picture list in which reference pictures are stored. When the refIdxLX value is 0, refIdxLX may indicate the first reference picture in the LX reference picture list. When the refIdxLX value is -1, refIdxLX may indicate that the reference picture does not exist in the reference picture list. In addition, predFlagLX, which will be described later, may indicate whether motion compensation is performed when generating a prediction block. For example, when the predFlagLX value is 1, the encoder and the decoder may perform motion compensation when generating the prediction block.

If the input motion vectors mvL0Col and mvL1Col are the same and RefPicOrder(currPic, refIdxL0, L0) and RefPicOrder(currPic, refIdxL1, L1) are the same (that is, if the L0 reference picture number and the L1 reference picture number are the same), the encoder and The decoder may perform the following process.

If there is a left uppermost block (A(xP-1, yP)) adjacent to the left of the current block, the left uppermost block is not an intra-mode coded block, predFlagL0A is '1' and mvL0A is (0, 0), the encoder and the decoder may perform the following process. Here, A may indicate that predFlagL0A and mvL0A are information about the upper left block (A).

mvL1Col = mvL0A

Otherwise, if there is an upper leftmost block (B(xP, yP-1)) adjacent to the upper end of the current block, the upper leftmost block is not an intra mode coded block, predFlagL0B is '1' and mvL0B If is not (0,0), the encoder and the decoder may perform the following process. Here, B may indicate that predFlagL0B and mvL0B are information about the upper left block (B).

mvL1Col = mvL0B

Otherwise, if there is an upper-left corner block (E(xP-1, yP-1)) located in the upper-left corner outside the current block, the upper-leftmost block is not an intra-mode coded block, and predFlagL0E is If '1' and mvL0E is not (0,0), the encoder and the decoder may perform the following process. Here, E may indicate that predFlagLOE and mvL0E are information about the upper left block (E).

mvL1Col = mvL0E

At this time, if mvL0Col and mvL1Col are the same, the encoder and the decoder may perform the following process.

avilableFlagL1Col = 0

Then, the encoder and the decoder may perform the following process.

availableFlagCol = availableFlagL0Col || availableFlagL1Col

predFlagLXCol = availableFlagLXCol

Here, availableFlagCol may indicate whether temporal motion information is derived, and predFlagLXCol may indicate whether final temporal motion information for each of L0 temporal motion information and L1 temporal motion information exists.

8 is a block diagram schematically illustrating an embodiment of an inter prediction apparatus capable of performing a process of deriving temporal motion information according to the embodiment of FIG. 6 . The inter prediction apparatus according to the embodiment of FIG. 8 includes a temporal motion information determination unit 810 , a reconstructed neighboring block motion information determination unit 820 , a prediction direction information resetting and L0 motion information setting unit 830 , and L1 temporal motion information A reset unit 840 may be included.

Referring to FIG. 8 , the temporal motion information determining unit 810 determines whether L0 temporal motion information and L1 temporal motion information are the same in the input temporal motion information, that is, the L0 reference picture number and the L1 reference picture number are the same, and the L0 motion vector It can be determined whether and L1 motion vectors are the same.

When the L0 temporal motion information and the L1 temporal motion information are not the same, the inter prediction apparatus may use the input temporal motion information as the temporal motion information of the current block as it is. When AMVP is applied, the temporal motion information of the current block may be determined or registered as a prediction motion vector candidate for the current block. Also, when the merge is applied, the temporal motion information of the current block may be determined or registered as a merge candidate for the current block. When the L0 temporal motion information and the L1 temporal motion information are the same, a determination process by the restored neighboring block motion information determiner 820 may be performed.

On the other hand, as described above in FIG. 6 , the temporal motion information determining unit 810 determines whether the L0 temporal motion information and the L1 temporal motion information are identical, not whether the L0 reference picture number and the L1 reference picture number are identical or whether the collocated block is predicted. direction can be determined. For example, if the L0 reference picture number and the L1 reference picture number are not the same, the inter prediction apparatus may use the input temporal motion information as it is as the temporal motion information of the current block, and the L0 reference picture number and the L1 reference picture number are In the same case, a determination process by the restored neighboring block motion information determination unit 820 may be performed. As another example, when the prediction direction of the collocated block is bidirectional prediction, the inter prediction apparatus can use the input temporal motion information as it is as the temporal motion information of the current block, and when the prediction direction of the collocated block is unidirectional prediction, the reconstructed neighboring block motion A determination process by the information determination unit 820 may be performed.

The reconstructed neighboring block motion information determining unit 820 may determine whether motion information of the reconstructed neighboring block exists. When motion information of a reconstructed neighboring block does not exist (eg, when a block having motion information does not exist among reconstructed neighboring blocks at a predetermined position and/or a position selected by a predetermined method), prediction direction information The reset and L0 motion information setting unit 830 may perform a process of resetting prediction direction information and setting L0 motion information. In addition, when motion information of a reconstructed neighboring block exists (eg, when a block having motion information exists among reconstructed neighboring blocks at a predetermined position and/or a position selected by a predetermined method), L1 temporal motion A reset process by the information resetting unit 840 may be performed.

When motion information of the reconstructed neighboring block does not exist, the prediction direction information resetting and L0 motion information setting unit 830 may reset the prediction direction information of the current block to unidirectional prediction. Also, at this time, the prediction direction information resetting and L0 motion information setting unit 830 may set only the L0 temporal motion information among the input temporal motion information as the final temporal motion information of the current block.

When motion information of the reconstructed neighboring block exists, the L1 temporal motion information resetter 840 may reset the L1 temporal motion information of the current block to the motion information of the reconstructed neighboring block. That is, the inter prediction apparatus may use the motion information of the reconstructed neighboring block as the L1 temporal motion information of the current block. At this time, for example, when the L1 temporal motion information resetter 840 searches for motion information to be used as the L1 temporal motion information of the current block in the reconstructed neighboring block, the motion information having a zero vector (0, 0) is not selected. It is also possible to select only motion information that does not have a zero vector (0, 0). Meanwhile, the L1 temporal motion information resetting unit 840 may reset the L1 temporal motion information of the current block based on the motion information candidate list of the current block. A specific embodiment of each of the above-described methods has been described above with reference to FIGS. 6 and 7 , and thus will be omitted herein.

9 is a flowchart schematically illustrating another embodiment of a method for deriving temporal motion information of a current block according to the present invention.

Embodiments to be described below will be described based on temporal motion information, but the present invention is not limited thereto. For example, the methods according to the embodiment of FIG. 9 include not only temporal motion information in the merge mode and/or AMVP mode, but also motion information of a current block derived based on the merge candidate list in the merge mode and/or in the AMVP mode. The same or similar method may be applied to motion information of the current block derived based on the prediction motion vector candidate list.

As described above, the temporal motion information may be derived based on the motion information of the collocated block corresponding to the current block in the already reconstructed collocated picture. Here, the collocated picture may correspond to, for example, one picture among reference pictures included in the reference picture list. The encoder and the decoder may determine a predetermined relative position based on a block existing at the same spatial position as the current block in the collocated picture, and the determined relative position (eg, spatially identical to the current block). The collocated block may be derived based on the position inside and/or outside the block existing in the position). The temporal motion information derived based on the collocated block may include prediction direction information, L0 reference picture number, L1 reference picture number, L0 motion vector, L1 motion vector, and the like.

Referring to FIG. 9 , the encoder and the decoder determine whether L0 temporal motion information and L1 temporal motion information are the same in the temporal motion information derived based on the collocated block, that is, the L0 reference picture number and the L1 reference picture number are the same and the L0 motion information is the same. It may be determined whether the vector and the L1 motion vector are the same (S910). Hereinafter, only for the embodiments of FIGS. 9 to 11, which will be described later, for convenience of explanation, the temporal motion information input to step S910 before resetting the temporal motion information is input temporal motion information (L0 input temporal motion information, L1 input temporal motion information). do. In addition, the motion vector included in the input temporal motion information is an input motion vector (L0 input motion vector, L1 input motion vector), and the reference picture index included in the input temporal motion information is an input reference picture index (L0 input reference picture index, L1 input). A reference picture index) and a reference picture number included in the input temporal motion information are referred to as an input reference picture number (L0 input reference picture number, L1 input reference picture number).

When the L0 input temporal motion information and the L1 input temporal motion information are not the same, that is, the L0 input reference picture number and the L1 input reference picture number are not the same and/or the L0 input motion vector and the L1 input motion vector are not the same , the encoder and the decoder may use the input temporal motion information derived based on the collocated block as the temporal motion information of the current block as it is. When AMVP is applied, the temporal motion information of the current block may be determined or registered as a prediction motion vector candidate of the current block. Also, when the merge is applied, the temporal motion information of the current block may be determined or registered as a merge candidate of the current block.

When the L0 input temporal motion information and the L1 input temporal motion information are the same, the encoder and the decoder may determine whether motion information of the collocated block exists ( S920 ). Here, the collocated block may be a newly derived collocated block rather than a collocated block used for deriving input temporal motion information. For example, the encoder and the decoder may determine whether a block having motion information exists among collocated blocks at a predetermined position and/or a position selected by a predetermined method.

Hereinafter, only in the embodiments of FIGS. 9 to 11 , for convenience of explanation, the collocated block used for deriving input temporal motion information is referred to as a first collocated block, and as in step S920, it is a target for determining whether motion information exists and will be described later. As in step S940, the collocated block used for resetting the L1 input temporal motion information is referred to as a second collocated block. Also, the collocated picture including the first collocated block is referred to as a first collocated picture, and the collocated picture including the second collocated block is referred to as a second collocated picture. Here, for example, the first collocated picture and the second collocated picture may each correspond to one picture from among the reference pictures included in the reference picture list.

When motion information of the second collocated block does not exist (eg, when a block having motion information does not exist among the second collocated blocks at a predetermined position and/or a position selected by a predetermined method), the encoder and the decoder The group may reset the prediction direction information of the current block to unidirectional prediction (S930). Also, at this time, the encoder and the decoder may use only the L0 input temporal motion information as the temporal motion information of the current block.

When motion information of the second collocated block exists (for example, when a block having motion information exists among the second collocated blocks at a predetermined position and/or a position selected by a predetermined method), the encoder and the decoder The motion information of the second collocated block may be used as the final L1 temporal motion information of the current block (S940). That is, at this time, the encoder and the decoder may reset the L1 input temporal motion information of the current block to the motion information of the second collocated block. A specific embodiment of the second collocated picture and the second collocated block used for resetting the L1 input temporal motion information will be described later.

In the above-described embodiment, the method for deriving temporal motion information of the current block is described based on a flowchart as a series of steps, but one or more steps in the flowchart may be deleted. For example, in the above-described embodiment, steps S920 and S940 may be omitted. In this case, when the L0 input temporal motion information and the L1 input temporal motion information are the same, the encoder and the decoder may reset the prediction direction information of the current block to the unidirectional prediction (S930). Also, at this time, the encoder and the decoder may use only the L0 input temporal motion information as the final temporal motion information of the current block.

Meanwhile, in the above-described embodiment, it is determined whether to perform the processes S920 to S940 based on the sameness of the L0 input temporal motion information and the L1 input temporal motion information, but the encoder and the decoder perform the processes S920 to S940 based on different conditions. may decide

As an embodiment, the encoder and the decoder may determine whether to perform steps S920 to S940 based on the sameness of the L0 input reference picture number and the L1 input reference picture number. For example, when the L0 input reference picture number and the L1 input reference picture number are the same, the encoder and the decoder may perform processes S920 to S940.

As another embodiment, the encoder and the decoder may determine whether to perform steps S920 to S940 based on the prediction direction of the first collocated block. As described above, the prediction direction information may refer to information indicating whether unidirectional prediction or bidirectional prediction is applied to a block on which prediction is performed. Accordingly, the prediction direction may correspond to unidirectional prediction or bidirectional prediction. For example, when the motion information (prediction direction) of the first collocated block is unidirectional prediction rather than bidirectional prediction, the encoder and the decoder may perform processes S920 to S940. This is because, in the case where the prediction direction of the first collocated block is unidirectional prediction, as a result, in the input temporal motion information derived from the first collocated block, the L0 input temporal motion information and the L1 input temporal motion information may be the same.

As another embodiment, the encoder and the decoder may determine whether to perform steps S920 to S940 based on information on whether motion information exists in the first collocated block. For example, the encoder and the decoder may perform steps S920 to S940 when motion information does not exist in the first collocated block. In this case, in step S940 described above, the L0 input temporal motion information rather than the L1 input temporal motion information may be reset. That is, the encoder and the decoder may set the L0 input temporal motion information as the motion information of the second collocated block, and may perform unidirectional prediction instead of bidirectional prediction on the current block. Also, when there is no motion information in the first collocated block, the encoder and the decoder may reset both the L0 input temporal motion information and the L1 input temporal motion information in step S940. That is, the encoder and the decoder may reset both the L0 input temporal motion information and the L1 input temporal motion information to the motion information of the second collocated block, and may perform bidirectional prediction on the current block.

As another embodiment, the encoder and the decoder may determine whether to perform steps S920 to S940 based on the size of the current block. For example, the encoder and the decoder may determine whether the size of the current block is smaller than a predetermined size. Here, the current block may be a CU, a PU, and/or a TU, and the predetermined size may be, for example, one of 8x8, 16x16, or 32x32. In this case, when the size of the current block is smaller than a predetermined size, the encoder and the decoder may perform processes S920 to S940.

As another embodiment, the encoder and the decoder may perform steps S920 to S940 when the L0 motion vector and/or the L1 motion vector in the motion information of the first collocated block corresponds to a zero vector (0,0). In this case, in step S940 described above, the encoder and the decoder may reset the motion vector(s) corresponding to the zero vector (0,0). As an example, the motion vector(s) corresponding to the zero vector (0,0) may be set as the motion vector of the second collocated block, and as another example, the motion vector(s) corresponding to the zero vector (0,0) ( ) may be set as a motion vector of a reconstructed neighboring block. As another example, the motion vector(s) corresponding to the zero vector (0,0) is a motion vector of a block located near the first collocated block. may be set. As another embodiment, the encoder and the decoder may perform steps S920 to S940 when the L0 motion vector and/or the L1 motion vector in the motion information of the first collocated block do not correspond to the zero vector (0,0). . In this case, in step S940 described above, the encoder and the decoder may reset motion vector(s) that do not correspond to the zero vector (0,0), and the motion vector(s) that do not correspond to the zero vector (0,0) ) may be reset to the motion vector of the second collocated block.

The conditions for determining whether to perform the processes S920 to S940 are not limited to the above-described embodiments, and various conditions may be applied according to conditions and/or needs.

In the above-described embodiments, since the L0 input temporal motion information and the L1 input temporal motion information are temporally derived motion information, there is a high possibility that they correspond to motion information due to the movement of an object. Therefore, when the encoder and the decoder search for motion information to be used as the final L1 temporal motion information of the current block in the second collocated block, motion information having a zero vector (0,0) is not selected and the zero vector (0,0) is not selected. It is also possible to select motion information that does not have . This is because a block having motion information corresponding to a zero vector (0, 0) is highly likely to correspond to a background rather than an object.

Meanwhile, the reset final L1 temporal motion information may be the same as the L0 input temporal motion information of the current block. Accordingly, when searching for motion information to be used as the final L1 temporal motion information of the current block in the second collocated block, the encoder and the decoder may select motion information that is not the same as the L0 input temporal motion information. For example, as in S940, when the final L1 temporal motion information of the current block is derived based on the second collocated block, the encoder and the decoder refer to a block having motion information different from the L0 input temporal motion information of the current block. It can be determined as a block used for deriving the final L1 temporal motion information. In this case, the encoder and the decoder may select only motion information whose difference from the L0 input temporal motion information of the current block is less than or equal to a predetermined threshold as motion information to be used as the final L1 temporal motion information. Here, the predetermined threshold value may be determined based on mode information of the current block, motion information of the current block, mode information of the neighboring block, and/or motion information of the neighboring block, and may be determined in various ways.

*In addition, in the above-described embodiments, the motion information of the second collocated block may include both L0 motion information and L1 motion information. In this case, the encoder and the decoder may determine one of the L0 motion information and the L1 motion information as motion information to be used as the final L1 temporal motion information of the current block. As an example, the encoder and the decoder may use the L0 motion information of the second collocated block as the final L1 temporal motion information of the current block. In this case, when the L0 motion information does not exist in the second collocated block, the encoder and the decoder may use the L1 motion information of the second collocated block as the final L1 temporal motion information for the current block. As another example, the encoder and the decoder may use the L1 motion information of the second collocated block as the final L1 temporal motion information of the current block. In this case, when the L1 motion information does not exist in the second collocated block, the encoder and the decoder may use the L0 motion information of the second collocated block as the final L1 temporal motion information for the current block.

Meanwhile, the encoder and the decoder may not use the motion information of the second collocated block to reset the L1 input temporal motion information (eg, the L1 input motion vector) of the current block in step S940 described above. In this case, the encoder and the decoder may reset L1 input temporal motion information (eg, L1 input motion vector) of the current block based on the motion information (eg, motion vector) of the first collocated block. Hereinafter, related embodiments will be described, which will be described based on a motion vector.

In one embodiment, the encoder and the decoder are motion indicating a relative position moved based on a predetermined distance and/or direction from a position indicated by motion information (motion vector) of the first collocated block within the first collocated picture. The information may be used as the final L1 temporal motion information (L1 motion vector) of the current block. As an example, the encoder and the decoder have a 1/4 pixel size (eg, (-1,0), (1,0), ( 0,-1), (0,1), (-1,-1), (-1,1), (1,-1), (1,1), etc. where 1/4 pixel unit is 1 .) may be used as the final L1 temporal motion information of the current block, the motion information indicating the moved position. As another example, the encoder and the decoder have a 1/2 pixel size (eg, (-2,0), (2,0), ( 0,-2), (0,2), (-2,-2), (-2,2), (2,-2), (2,2), etc. where 1/4 pixel unit is 1 .) may be used as the final L1 temporal motion information of the current block, the motion information indicating the moved position. As another example, the encoder and the decoder have integer pixel sizes (eg, (-4,0), (4,0), (0) in the vertical and/or horizontal direction at the position indicated by the motion information of the first collocated block. ,-4), (0,4), (-4,-4), (-4,4), (4,-4), (4,4), etc. where 1/4 pixel unit is 1 The motion information indicating the position moved as much as applicable.) may be used as the final L1 temporal motion information of the current block. On the other hand, since the motion vector includes a vertical component and a horizontal component, the above-described methods (moving by 1/4 pixel size, moving by 1/2 pixel size, moving by integer pixel size) include a horizontal component and a vertical component. Each component may be independently applied. In this case, the moving distance in the horizontal direction and the moving distance in the vertical direction may be different from each other.

In another embodiment, the encoder and the decoder may change the value of the motion information (motion vector) of the first collocated block to a value of another pixel unit, and then use the changed value as the final L1 temporal motion information (L1 motion vector) of the current block. have. For example, when the motion information value of the first collocated block is a value in units of 1/4 pixels, the encoder and the decoder change the value of the motion information of the first collocated block to a value in units of 1/2 pixels based on a shift operation, etc. and the changed value may be used as the final L1 temporal motion information of the current block. As another example, when the motion information value of the first collocated block is a value in units of 1/2 pixels, the encoder and the decoder may change the value of the motion information of the first collocated block to a value in units of integer pixels based on a shift operation, etc. , the changed value may be used as the final L1 temporal motion information of the current block.

The method of resetting the L1 input temporal motion information (eg, L1 input motion vector) of the current block based on the motion information (eg, motion vector) of the first collocated block is not limited to the above-described embodiment, It may be applied in various forms according to implementation and/or needs. In addition, the methods according to the above-described embodiment may be applied in the same or similar manner to the case of resetting the L1 input temporal motion information based on the motion information of the second collocated block as in step S940 of FIG. 9 .

Meanwhile, in the above-described embodiment, the input temporal motion information input in step S910 may include an input reference picture index as well as an input motion vector. Here, the L0 input motion vector and the L1 input motion vector may be motion vectors temporally derived based on the first collocated block as described above, and the L0 input reference picture index and the L1 input reference picture index are obtained from the reconstructed neighboring blocks. It may be a spatially derived reference picture index. In this case, the L0 input reference picture index and the L1 input reference picture index may be set to the smallest non-negative value among the reference picture indexes of the reconstructed neighboring blocks. Meanwhile, as another example, the L0 input reference picture index and the L1 input reference picture index may be set to 0 regardless of the motion information of the reconstructed neighboring blocks.

When the L1 input motion vector is reset based on the second collocated block, a reset process may also be performed on the L1 input reference picture index. Hereinafter, embodiments of the reset process for the L1 input reference picture index will be described.

As an embodiment, the encoder and the decoder may reset the L1 input reference picture index based on the second collocated block. As described above, the encoder and the decoder may use the motion information of the second collocated block as the final L1 temporal motion information of the current block. In this case, the encoder and the decoder may derive the final L1 reference picture index by resetting the L1 input reference picture index to the reference picture index of the second collocated block. In another embodiment, the encoder and the decoder may derive the final L1 reference picture index by resetting the L1 input reference picture index value to a predetermined fixed reference picture index value.

In another embodiment, L0 input temporal motion information (eg, L0 input motion vector, L0 input reference picture index, etc.) and L1 input temporal motion information (eg, L1 input motion vector, L1 input reference picture index, etc.) ) is the same, the encoder and the decoder may reset both the L0 input reference picture index and the L1 input reference picture index to a value of 0 to use as final temporal motion information. This is because, when the L0 input temporal motion information and the L1 input temporal motion information are the same, there is a high probability that both the L0 reference picture index and the L1 reference picture index are 0.

As another embodiment, the encoder and the decoder may reset the L1 input reference picture index value to the same value as the L0 input reference picture index value. On the other hand, the encoder and the decoder may reset the L1 input reference picture index value to a reference picture index value that is not the same as the L0 input reference picture index. Among the reference picture indexes of the reconstructed neighboring blocks, there may be a reference picture index that is not the same as the L0 input reference picture index. In this case, for example, the encoder and the decoder may use the most frequently used reference picture index among reference picture indexes that are not the same as the L0 input reference picture index to reset the L1 input reference picture index. As another example, in this case, the encoder and the decoder may use the reference picture index having the smallest non-negative value among the reference picture indices that are not the same as the L0 input reference picture index to reset the L1 input reference picture index.

Meanwhile, as described above, the encoder and the decoder may derive the final L1 temporal motion vector by resetting the L1 input motion vector value to the same value as the motion vector of the second collocated block. In this case, the motion vector of the second collocated block may be scaled and used according to the L1 input reference picture index and/or the reset L1 reference picture index. The L1 input reference picture index may be used as the final L1 reference picture index as it is without a reset process, or may be used as the final L1 reference picture index through the reset process as in the above-described embodiment. In this case, the reference picture corresponding to the motion vector of the second collocated block and the reference picture indicated by the final L1 reference picture index may be different from each other. In this case, the encoder and the decoder may perform scaling on the motion vector of the second collocated block and use the scaled motion vector as the final L1 temporal motion vector of the current block.

The above-described embodiments may be combined and applied in various ways according to a motion vector resetting process and a reference picture index resetting process (eg, RefIdxLX, X=0,1). Hereinafter, it is assumed that the L1 input motion vector is reset based on the second collocated block in embodiments to be described below. That is, it is assumed that the encoder and the decoder reset the L1 input motion vector to one of the motion vectors of the second collocated block.

As an embodiment, the encoder and the decoder may use the first collocated picture used to derive the L0 input temporal motion information as it is as the second collocated picture, and the second collocated block may be derived from the second collocated picture. In this case, the encoder and the decoder may scale and use the motion vector of the second collocated block to reset the L1 input motion vector. In this case, the scaled motion vector may be used as the final L1 temporal motion information. In this case, for example, the L1 input reference picture index may be used as the final L1 temporal motion information as it is without a reset process. That is, the encoder and the decoder may allocate the L1 input reference picture index to the L1 temporal motion information of the current block as it is. As another example, the L1 input reference picture index value may be reset to a predetermined fixed reference picture index value. That is, the encoder and the decoder may allocate a predetermined fixed reference picture index value to the L1 temporal motion information of the current block, and the predetermined fixed reference picture index may be used as the final L1 temporal motion information.

In another embodiment, the encoder and the decoder may use the first collocated picture used for deriving the L0 input temporal motion information as it is as the second collocated picture as in the above-described embodiment, and the second collocated block is the second collocated picture. can be derived from At this time, in order to reset the L1 input motion vector, the encoder and the decoder derive a motion vector that is not the same as the L0 input motion vector from among the motion vectors of the second collocated block and has a difference from the L0 input motion vector equal to or less than a predetermined threshold. can do. The encoder and the decoder may perform scaling on the derived motion vector, and the scaled motion vector may be used as final L1 temporal motion information. In this case, for example, the L1 input reference picture index may be used as the final L1 temporal motion information as it is without a reset process. That is, the encoder and the decoder may allocate the L1 input reference picture index to the L1 temporal motion information of the current block as it is. As another example, the L1 input reference picture index value may be reset to a predetermined fixed reference picture index value. That is, the encoder and the decoder may allocate a predetermined fixed reference picture index value to the L1 temporal motion information of the current block, and the predetermined fixed reference picture index may be used as the final L1 temporal motion information.

As another embodiment, the encoder and the decoder may use a reference picture that is not the same as the first collocated picture used for deriving the L0 input temporal motion information as the second collocated picture. Here, the second collocated picture may be, for example, a most recently decoded reference picture among reference pictures in the L1 reference picture list that is not identical to the first collocated picture. As another example, the encoder and the decoder may select the most frequently used reference picture as the second collocated picture based on the reference picture numbers of the reconstructed neighboring blocks. The second collocated block may be derived from the second collocated picture, and in this case, the encoder and the decoder may scale and use the motion vector of the second collocated block to reset the L1 input motion vector. In this case, the scaled motion vector may be used as the final L1 temporal motion information. In this case, for example, the L1 input reference picture index may be used as the final L1 temporal motion information as it is without a reset process. That is, the encoder and the decoder may allocate the L1 input reference picture index to the L1 temporal motion information of the current block as it is. As another example, the L1 input reference picture index value may be reset to a predetermined fixed reference picture index value. That is, the encoder and the decoder may allocate a predetermined fixed reference picture index value to the L1 temporal motion information of the current block, and the predetermined fixed reference picture index may be used as the final L1 temporal motion information.

As another embodiment, the encoder and the decoder may use, as the second collocated picture, a reference picture that is not the same as the first collocated picture used for deriving the L0 input temporal motion information as in the above-described embodiment, and the second collocated block may be derived from the second collocated picture. Here, the second collocated picture may be, for example, a most recently decoded reference picture among reference pictures in the L1 reference picture list that is not identical to the first collocated picture. As another example, the encoder and the decoder may select the most frequently used reference picture as the second collocated picture based on the reference picture numbers of the reconstructed neighboring blocks. At this time, in order to reset the L1 input motion vector, the encoder and the decoder derive a motion vector that is not the same as the L0 input motion vector from among the motion vectors of the second collocated block and has a difference from the L0 input motion vector equal to or less than a predetermined threshold. can do. The encoder and the decoder may perform scaling on the derived motion vector, and the scaled motion vector may be used as final L1 temporal motion information. In this case, for example, the L1 input reference picture index may be used as the final L1 temporal motion information as it is without a reset process. That is, the encoder and the decoder may allocate the L1 input reference picture index to the L1 temporal motion information of the current block as it is. As another example, the L1 input reference picture index value may be reset to a predetermined fixed reference picture index value. That is, the encoder and the decoder may allocate a predetermined fixed reference picture index value to the L1 temporal motion information of the current block, and the predetermined fixed reference picture index may be used as the final L1 temporal motion information.

The combination of the embodiments of the motion vector resetting process and the reference picture index resetting process is not limited to the above-described embodiment, and various types of combinations may be provided as well as the above-described embodiments according to implementation and/or necessity. have.

Meanwhile, the reset L1 temporal motion information may be the same as the L0 temporal motion information of the current block. Accordingly, when the reset L1 temporal motion information is the same as the L0 temporal motion information of the current block, the encoder and the decoder may reset the prediction direction information of the current block to unidirectional prediction. In this case, the encoder and the decoder may use only the L0 temporal motion information as the temporal motion information of the current block. This method may be applied to the present invention as a combination combined with the above-described embodiments.

10 is a diagram schematically illustrating an embodiment of a second collocated block used for resetting L1 temporal motion information.

The encoder and the decoder may determine the second collocated block based on a co-located block that is spatially co-located with the current block in the second collocated picture. In FIG. 10 , a block 1010 indicates a current block, and a block 1020 indicates a co-located block in the second collocated picture. Here, the second collocated picture may be determined in various ways.

As an embodiment, the encoder and the decoder may determine one reference picture from among reference pictures included in the L0 reference picture list as the second collocated picture. As an example, the encoder and the decoder may determine the first reference picture in the L0 reference picture list as the second collocated picture. As another example, the encoder and the decoder may determine the second reference picture in the L0 reference picture list as the second collocated picture. As another example, the encoder and the decoder may determine the third reference picture in the L0 reference picture list as the second collocated picture. As another example, the encoder and the decoder may determine the fourth reference picture in the L0 reference picture list as the second collocated picture.

In another embodiment, the encoder and the decoder may determine one reference picture from among reference pictures included in the L1 reference picture list as the second collocated picture. As an example, the encoder and the decoder may determine the first reference picture in the L1 reference picture list as the second collocated picture. As another example, the encoder and the decoder may determine the second reference picture in the L1 reference picture list as the second collocated picture. As another example, the encoder and the decoder may determine the third reference picture in the L1 reference picture list as the second collocated picture. As another example, the encoder and the decoder may determine the fourth reference picture in the L1 reference picture list as the second collocated picture.

In another embodiment, the encoder and/or decoder selects the reference picture that provides the highest coding efficiency among all reference pictures (and/or some reference pictures determined according to a predetermined condition) in the L0 reference picture list and the L1 reference picture list. It can be used as the second collocated picture. Here, the encoding efficiency may be determined based on motion information of a block existing at a position corresponding to the second collocated block in each reference picture. In this case, the encoder may derive the second collocated picture based on the encoding efficiency, and may transmit second collocated picture information indicating the second collocated picture to the decoder. The decoder may determine the second collocated picture based on the transmitted second collocated picture information.

In the above-described embodiments, the encoder and the decoder may determine only the reference picture that is not the same as the first collocated picture used for deriving the L0 input temporal motion information as the second collocated picture. In this case, only the reference picture that is not the same as the first collocated picture may be used to derive the final L1 temporal motion information of the current block.

The method of determining the second collocated picture is not limited to the above-described embodiment, and the second collocated picture may be determined in another manner according to implementation and/or necessity.

On the other hand, in the embodiment of Figure 10, the block located at the center inside the co-located block 1020 is block D, the block located at the upper left corner inside the co-located block 1020 is block S, the co-located block 1020 inside A block located at the lower left corner of the block Q is referred to as block Q, a block located at the upper right corner inside the co-located block 1020 is referred to as block R, and a block located at the lower right corner inside the co-located block 1020 is referred to as block C. In addition, among the blocks adjacent to the left of the co-located block 1020 , the uppermost block is block F, the lowest-located block among the blocks adjacent to the left of the co-located block 1020 is block J, and the co-located block 1020 ). The leftmost block among the blocks adjacent to the top of The block located at is block N, the lowermost block among the blocks adjacent to the right of the same location block 1020 is block B, the leftmost block among the blocks adjacent to the bottom of the same location block 1020 is block K, the same Among the blocks adjacent to the lower end of the location block 1020 , the rightmost block is referred to as a block A. In addition, the block located at the upper left corner outside the co-located block 1020 is block E, the block located at the lower left corner outside the co-located block 1020 is block I, the co-located block 1020 is outside the upper right corner The located block is referred to as block L, and the block located at the lower right corner outside the same location block 1020 is referred to as block H. In addition, a block located adjacent to the right side of block B is referred to as block P, and a block located adjacent to the lower end of block A is referred to as block O.

As described above in the embodiment of FIG. 9 , the encoder and the decoder may use the motion information of the second collocated block as the final L1 temporal motion information of the current block according to a predetermined condition. That is, the encoder and the decoder may reset the L1 input temporal motion information value to the motion information value of the second collocated block. In this case, the motion used as the second collocated block and/or the final L1 temporal motion information may be derived by various methods.

In one embodiment, the encoder and the decoder derive motion information used as the final L1 temporal motion information from one block existing at a predetermined position with respect to the same position block among blocks located inside and/or outside the same position block. can do. In this case, one block existing at the predetermined position may correspond to the second collocated block. Here, the block at the predetermined position is a block A, a block B, a block C, a block D, a block E, a block F, a block G, a block H, a block I, a block J, a block K, a block L, a block M, a block N , block O, block P, block Q, block R, or block S.

In another embodiment, the encoder and the decoder may scan a plurality of blocks existing at a predetermined position among blocks located inside and/or outside the same position block in a predetermined order. In this case, the encoder and the decoder may use the motion information of the first block in which the motion information exists in the scan order as the final L1 temporal motion information of the current block. Here, the first block in which the motion information exists may correspond to the second collocated block. A scan target block and a scan order may be determined in various forms. For example, the encoder and the decoder may scan the blocks in the order of block H, block D, block A, block B, block C, block I, block J, block F, block G, and block E.

As another embodiment, the encoder and the decoder may derive an intermediate value with respect to motion information of a plurality of blocks existing at a predetermined position among blocks located inside and/or outside the same position block, and the derived intermediate value can be used as the final L1 temporal motion information of the current block. For example, the plurality of blocks may be block H, block D, block A, block B, block C, block I, block J, block F, block G, and block E.

A method of deriving the L1 temporal motion information of the current block from the second collocated block is not limited to the above-described embodiment, and the L1 temporal motion information of the current block may be derived in various ways according to implementation and/or necessity.

11 is a block diagram schematically illustrating an embodiment of an inter prediction apparatus capable of performing a process of deriving temporal motion information according to the embodiment of FIG. 9 . The inter prediction apparatus according to the embodiment of FIG. 11 includes a temporal motion information determination unit 1110 , a second collocated block motion information determination unit 1120 , a prediction direction information resetting and L0 motion information setting unit 1130 , and L1 temporal motion information A reset unit 1140 may be included.

Referring to FIG. 11 , the temporal motion information determining unit 1110 determines whether L0 input temporal motion information and L1 input temporal motion information are the same in the input temporal motion information, that is, the L0 input reference picture number and the L1 input reference picture number are the same. and it may be determined whether the L0 input motion vector and the L1 input motion vector are the same.

When the L0 input temporal motion information and the L1 input temporal motion information are not the same, the inter prediction apparatus may use the input temporal motion information as the temporal motion information of the current block as it is. When AMVP is applied, the temporal motion information of the current block may be determined or registered as a prediction motion vector candidate for the current block. Also, when the merge is applied, the temporal motion information of the current block may be determined or registered as a merge candidate for the current block. When the L0 input temporal motion information and the L1 input temporal motion information are the same, a determination process by the second collocated block motion information determiner 1120 may be performed.

On the other hand, as described above in FIG. 9 , the temporal motion information determining unit 1110 determines whether the L0 input temporal motion information and the L1 input temporal motion information are identical, not whether the L0 input reference picture number and the L1 input reference picture number are identical or not. The prediction direction of the first collocated block may be determined. For example, when the L0 input reference picture number and the L1 input reference picture number are not the same, the inter prediction apparatus may use the input temporal motion information as it is as the temporal motion information of the current block, and the L0 input reference picture number and L1 input When the reference picture numbers are the same, a determination process by the second collocated block motion information determination unit 1120 may be performed. As another example, when the prediction direction of the first collocated block is bidirectional prediction, the inter prediction apparatus may use the input temporal motion information as it is as the temporal motion information of the current block, and when the prediction direction of the first collocated block is unidirectional prediction, the inter prediction apparatus 2 A determination process by the collocated block motion information determination unit 1120 may be performed.

The second collocated block motion information determiner 1120 may determine whether motion information of the second collocated block exists. When motion information of the second collocated block does not exist (eg, when a block having motion information does not exist among the second collocated blocks at a predetermined position and/or a position selected by a predetermined method), prediction direction information The resetting and L0 motion information setting unit 1130 may perform a process of resetting prediction direction information and setting L0 motion information. In addition, when motion information of the second collocated block exists (eg, when a block having motion information exists among second collocated blocks at a predetermined position and/or a position selected by a predetermined method), L1 temporal movement A resetting process by the information resetting unit 1140 may be performed.

When the motion information of the second collocated block does not exist, the prediction direction information resetting and L0 motion information setting unit 1130 may reset the prediction direction information of the current block to unidirectional prediction. Also, at this time, the prediction direction information resetting and L0 motion information setting unit 1130 may set only the L0 input temporal motion information among the input temporal motion information as the final temporal motion information of the current block.

When the motion information of the second collocated block exists, the L1 temporal motion information resetting unit 1140 may reset the L1 input temporal motion information of the current block to the motion information of the second collocated block. That is, the inter prediction apparatus may use the motion information of the second collocated block as the final L1 temporal motion information of the current block. At this time, when the motion information of the second collocated block includes a zero vector (0,0), the L1 temporal motion information resetter 1140 converts the motion information of a block located in the vicinity of the second collocated block to the last of the current block. It can also be used as L1 temporal motion information. A specific embodiment of each of the above-described methods has been described above with reference to FIGS. 6 and 7 , and thus will be omitted herein.

12 is a diagram schematically illustrating an embodiment of a method for deriving motion information of a current block according to the present invention.

As described above, the encoder and the decoder may derive the motion information of the current block based on the motion information of the reconstructed neighboring block and/or the motion information of the collocated block. Here, the reconstructed neighboring block is a block in the reconstructed current picture after the reconstructed neighboring block has already been encoded and/or decoded, and may include a block adjacent to the current block and/or a block located at an outer corner of the current block. Also, the collocated block may mean a block corresponding to the current block in the collocated picture, and the collocated picture may correspond to, for example, one picture among reference pictures in the reference picture list. In this case, motion information derived from the reconstructed neighboring block may be referred to as spatial motion information, and motion information derived based on the collocated block may be referred to as temporal motion information. Here, the spatial motion information and the temporal motion information may include prediction direction information, L0 reference picture number, L1 reference picture number, L0 motion vector, and L1 motion vector, respectively.

Meanwhile, in the video decoding process, an undecoded picture may exist among previous pictures of the current picture (and/or the picture to be decoded) due to excessive network traffic. In this case, an incorrect collocated block may be used in the process of deriving temporal motion information for a block in the current picture, and since accurate temporal motion information may not be derived, the current picture may not be properly decoded. Therefore, in order to solve the above problem, the encoder and the decoder, when deriving the motion information of the current block, one of the L0 motion information and the L1 motion information is spatially derived based on the reconstructed neighboring block, and the other is a collocated block. can be derived over time. That is, the encoder and the decoder may independently set the L0 motion information and the L1 motion information, respectively. In this case, even when an undecoded previous picture exists, the encoder and the decoder may allow the current picture to be reconstructed to some extent.

Referring to FIG. 12 , the encoder and the decoder may set L0 motion information ( 1210 ). In this case, the encoder and the decoder use motion information (spatial motion information) spatially derived based on the reconstructed neighboring block as the L0 motion information, or motion information derived temporally based on the collocated block (temporal motion information) may be used as L0 motion information. That is, the L0 motion information may be spatially derived based on the reconstructed neighboring block or temporally based on the collocated block.

Referring back to FIG. 12 , the encoder and the decoder may set L1 motion information ( 1220 ). At this time, the encoder and the decoder use motion information (spatial motion information) spatially derived based on the reconstructed neighboring block as L1 motion information, or motion information derived temporally based on the collocated block (temporal motion information) may be used as L1 motion information. That is, the L1 motion information may be spatially derived based on a reconstructed neighboring block or temporally based on a collocated block.

As an embodiment, when the L0 motion information is motion information temporally derived based on the collocated block, the encoder and the decoder may use the spatial motion information derived based on the reconstructed neighboring block as the L1 motion information. As another embodiment, when the L0 motion information is motion information spatially derived based on the reconstructed neighboring block, the encoder and the decoder may use the temporal motion information derived based on the collocated block as the L1 motion information.

Meanwhile, in the above-described embodiment, the encoder and the decoder may derive the L1 motion information based on the L0 motion information. As an example, it is assumed that the L0 motion information is motion information derived based on a collocated block. In this case, as described above, the encoder and the decoder may use spatial motion information spatially derived based on the reconstructed neighboring blocks as the L1 motion information. In this case, for example, the encoder and the decoder may derive only motion information identical to or similar to the L0 motion information as spatial motion information to be used as the L1 motion information. As another example, the encoder and the decoder may derive only motion information that is not identical to the L0 motion information as spatial motion information to be used as the L1 motion information. In this case, the encoder and the decoder may derive only motion information whose difference from the L0 motion information is less than or equal to a predetermined threshold as spatial motion information to be used as the L1 motion information. Here, the predetermined threshold value may be determined based on mode information of the current block, motion information of the current block, mode information of the neighboring block, and/or motion information of the neighboring block, and may be determined in various ways.

A method of setting the L0 motion information and the L1 motion information is not limited to the above-described embodiment, and may vary according to implementation and/or necessity.

On the other hand, when the L0 motion information and/or the L1 motion information is spatially derived based on the motion information of the reconstructed neighboring block, the motion information of the reconstructed neighboring block is the spatially derived spatial motion information and the temporally derived temporal motion. It may contain all information. In this case, the encoder and the decoder may derive the L0 motion information of the current block and/or the L1 motion information of the current block by using only spatially derived spatial motion information among the motion information of the reconstructed neighboring blocks.

Referring back to FIG. 12 , the encoder and the decoder may derive the motion information of the current block by integrating the L0 motion information and the L1 motion information ( 1230 ).

Meanwhile, in the derived motion information of the current block, the L0 motion information and the L1 motion information may be identical to each other. Accordingly, when the L0 motion information and the L1 motion information are the same, the encoder and the decoder may reset the prediction direction information of the current block to unidirectional prediction. In this case, the encoder and the decoder may use only the L0 motion information as the motion information of the current block.

13 is a block diagram schematically illustrating an embodiment of an inter prediction apparatus capable of performing a motion information derivation process according to the embodiment of FIG. 12 . The inter prediction apparatus according to the embodiment of FIG. 13 may include an L0 motion information setting unit 1310 , an L1 motion information setting unit 1320 , and a motion information integrator 1330 .

As described above in the embodiment of FIG. 12 , the encoder and the decoder may independently set the L0 motion information and the L1 motion information, respectively. In this case, the encoder and the decoder may spatially derive one of the L0 motion information and the L1 motion information based on the reconstructed neighboring block and the other temporally derive the other one based on the collocated block.

Referring to FIG. 13 , the L0 motion information setting unit 1310 may set L0 motion information. In this case, the encoder and the decoder use motion information (spatial motion information) spatially derived based on the reconstructed neighboring block as the L0 motion information, or motion information derived temporally based on the collocated block (temporal motion information) may be used as L0 motion information. That is, the L0 motion information may be spatially derived based on the reconstructed neighboring block or temporally based on the collocated block.

Referring back to FIG. 13 , the L1 motion information setting unit 1320 may set L1 motion information. At this time, the encoder and the decoder use motion information (spatial motion information) spatially derived based on the reconstructed neighboring block as L1 motion information, or motion information derived temporally based on the collocated block (temporal motion information) may be used as L1 motion information. That is, the L1 motion information may be spatially derived based on a reconstructed neighboring block or temporally based on a collocated block.

A specific embodiment of a method of setting the L0 motion information and the L1 motion information has been described above with reference to FIG. 12, and thus will be omitted herein.

Referring back to FIG. 13 , the motion information integrator 1330 integrates the L0 motion information set by the L0 motion information setting unit 1310 and the L1 motion information set by the L1 motion information setting unit 1320 , thereby moving the current block. information can be derived.

14 is a flowchart schematically illustrating another embodiment of a method for deriving temporal motion information of a current block according to the present invention.

Embodiments to be described below will be described based on temporal motion information, but the present invention is not limited thereto. For example, the methods according to the embodiment of FIG. 14 include not only temporal motion information in the merge mode and/or AMVP mode, but also motion information of a current block derived based on the merge candidate list in the merge mode and/or in the AMVP mode. The same or similar method may be applied to motion information of the current block derived based on the prediction motion vector candidate list.

As described above, the temporal motion information may be derived based on the motion information of the collocated block corresponding to the current block in the already reconstructed collocated picture. Here, the collocated picture may correspond to, for example, one picture among reference pictures included in the reference picture list. The encoder and the decoder may determine a predetermined relative position based on a block existing at the same spatial position as the current block in the collocated picture, and the determined relative position (eg, spatially identical to the current block). The collocated block may be derived based on the position inside and/or outside the block existing in the position). The temporal motion information derived based on the collocated block may include prediction direction information, L0 reference picture number, L1 reference picture number, L0 motion vector, L1 motion vector, and the like.

Referring to FIG. 14 , the encoder and the decoder determine whether L0 temporal motion information and L1 temporal motion information are the same in the temporal motion information derived based on the collocated block, that is, the L0 reference picture number and the L1 reference picture number are the same and the L0 motion information is the same. It may be determined whether the vector and the L1 motion vector are the same (S1410).

Hereinafter, only for the embodiments of FIGS. 14 to 15, which will be described later, for convenience of explanation, the temporal motion information input to step S1410 before resetting the temporal motion information is input temporal motion information (L0 input temporal motion information, L1 input temporal motion information). do. In this case, the input temporal motion information may correspond to temporal motion information derived based on the collocated block. In addition, the motion vector included in the input temporal motion information is an input motion vector (L0 input motion vector, L1 input motion vector), and the reference picture index included in the input temporal motion information is an input reference picture index (L0 input reference picture index, L1 input). A reference picture index) and a reference picture number included in the input temporal motion information are referred to as an input reference picture number (L0 input reference picture number, L1 input reference picture number). In addition, in an embodiment to be described later, the reference picture indicated by the L0 input reference picture number is referred to as an L0 reference picture, and the reference picture indicated by the L1 input reference picture number is referred to as an L1 reference picture.

When the L0 temporal motion information and the L1 temporal motion information are not the same, that is, the L0 reference picture number and the L1 reference picture number are not the same and/or the L0 motion vector and the L1 input motion vector are not the same, the encoder and the decoder The input temporal motion information derived based on the collocated block may be used as the final temporal motion information of the current block as it is. When AMVP is applied, the final temporal motion information may be determined or registered as a prediction motion vector candidate of the current block. Also, when merge is applied, the final temporal motion information may be determined or registered as a merge candidate of the current block.

When the L0 input temporal motion information and the L1 input temporal motion information are the same, the encoder and the decoder may search for or derive motion information to be used as the final L1 temporal motion information in the L1 reference picture (S1420). Here, since the L0 input temporal motion information and the L1 input temporal motion information are identical to each other, the L1 reference picture may correspond to the same picture as the L0 reference picture. Accordingly, in an embodiment to be described later, the L1 reference picture may mean the same picture as the L0 reference picture.

For example, the encoder and the decoder may search and derive motion information from the L1 reference picture in the same way. In this case, the encoder may not transmit the motion information derived from the L1 reference picture to the decoder. As another example, the encoder may derive motion information from the L1 reference picture by a predetermined method, then include the derived motion information in a bitstream and transmit it to the decoder. In this case, the decoder may determine motion information to be used as the final L1 temporal motion information based on the transmitted motion information. The motion information derived from the encoder may include a reference picture index and a motion vector, and the encoder may independently transmit the reference picture index and the motion vector to the decoder. In this case, the encoder may obtain a difference value between the actual motion vector of the current block and the motion vector derived from the L1 reference picture, and then transmit the difference value to the decoder.

Referring back to FIG. 14 , the encoder and the decoder may use the motion information derived from the L1 reference picture as the final L1 temporal motion information of the current block ( S1430 ). That is, at this time, the encoder and the decoder may reset the L1 input temporal motion information value to the value of the motion information derived from the L1 reference picture.

Hereinafter, embodiments of a method of deriving motion information to be used as final L1 temporal motion information from an L1 reference picture will be described.

In an embodiment, the encoder and the decoder generate motion information indicating a position within a predetermined range with respect to a position indicated by the L0 input temporal motion information (motion vector) in the L1 reference picture, and finally L1 temporal motion information. (L1 motion vector) can be used. Specifically, the predetermined range corresponds to a range including a position within a distance of +T and/or -T in the vertical and/or horizontal direction with respect to the position indicated by the L0 input temporal motion information in the L1 reference picture. can be Here, as an example, T may be a value indicating a distance in units of 1/4 pixels, and as another example, T may be a value indicating a distance in units of 1/2 pixels. As another example, T may be a value indicating a distance in units of integer pixels. When an integer pixel unit is used, since an interpolation process may not be performed during motion compensation, calculation complexity may be reduced.

As another example, the encoder may derive a region that best matches or is similar to an input block (original block) corresponding to the current block in the L1 reference picture. In this case, the encoder may derive motion information to be used as the final L1 temporal motion information of the current block based on the derived region. For example, the encoder may derive a motion vector to be used as the final L1 temporal motion information based on the position of the current block and the position of the block corresponding to the derived region, and based on the block corresponding to the derived region, the final A reference picture index (and/or reference picture number) to be used as the L1 temporal motion information may be derived. The derived motion information may be included in the bitstream and transmitted from the encoder to the decoder. In this case, the decoder may reset the L1 input temporal motion information based on the transmitted motion information.

The reference picture index (and/or reference picture number) derived from the L1 reference picture may be different from the L1 input reference picture index (and/or the L1 input reference picture number). In this case, the encoder and the decoder may use the reference picture index (and/or reference picture number) derived from the L1 reference picture as the final L1 temporal motion information of the current block.

A method of deriving motion information to be used as the final L1 temporal motion information based on the L1 reference picture is not limited to the above-described embodiment, and the encoder and the decoder may derive the motion information in other ways according to implementation and/or necessity. .

Meanwhile, in the above-described embodiment, it is determined whether to perform the processes S1420 to S1430 based on the sameness of the L0 input temporal motion information and the L1 input temporal motion information, but the encoder and the decoder perform the processes S1420 to S1430 based on different conditions. may decide

As an embodiment, the encoder and the decoder may determine whether to perform steps S1420 to S1430 based on the prediction direction of the collocated block. As described above, the prediction direction information may refer to information indicating whether unidirectional prediction or bidirectional prediction is applied to a block on which prediction is performed. Accordingly, the prediction direction may correspond to unidirectional prediction or bidirectional prediction. For example, when the motion information (prediction direction) of the collocated block is unidirectional prediction rather than bidirectional prediction, the encoder and the decoder may perform processes S1420 to S1430. This is because, in the case where the prediction direction of the first collocated block is unidirectional prediction, as a result, in the input temporal motion information derived from the first collocated block, the L0 input temporal motion information and the L1 input temporal motion information may be the same.

As another embodiment, the encoder and the decoder may determine whether to perform steps S1420 to S1430 based on information on whether motion information exists in the collocated block. For example, the encoder and the decoder may perform operations S1420 to S1430 when motion information does not exist in the collocated block. In this case, in step S1430 described above, the L0 input temporal motion information, not the L1 input temporal motion information, may be reset. That is, the encoder and the decoder may reset the L0 input temporal motion information to the motion information of the L1 reference picture, and may perform unidirectional prediction instead of bidirectional prediction on the current block. Also, when there is no motion information in the collocated block, the encoder and the decoder may reset both the L0 input temporal motion information and the L1 input temporal motion information in step S1430. That is, the encoder and the decoder may reset both the L0 input temporal motion information and the L1 input temporal motion information to the motion information of the L1 reference picture, and may perform bidirectional prediction on the current block.

As another embodiment, the encoder and the decoder may perform steps S1420 to S1430 when the L0 motion vector and/or the L1 motion vector in the motion information of the collocated block corresponds to a zero vector (0,0). In this case, in step S1430 described above, the encoder and the decoder may reset the motion vector(s) corresponding to the zero vector (0,0). As an example, the motion vector(s) corresponding to the zero vector (0,0) may be set as the motion vector of the L1 reference picture, and as another example, the motion vector(s) corresponding to the zero vector (0,0) ) may be set as a motion vector of a reconstructed neighboring block, and as another example, the motion vector(s) corresponding to the zero vector (0,0) may be set as a motion vector of a block located near the collocated block. have. As another embodiment, the encoder and the decoder may perform steps S1420 to S1430 when the L0 motion vector and/or the L1 motion vector in the motion information of the collocated block do not correspond to the zero vector (0,0). In this case, in step S1430, the encoder and the decoder may reset motion vector(s) that do not correspond to the zero vector (0,0), and the motion vector(s) that do not correspond to the zero vector (0,0) ) may be reset to the motion vector of the L1 reference picture.

Conditions for determining whether to perform processes S1420 to S1430 are not limited to the above-described embodiments, and various conditions may be applied according to conditions and/or needs.

Meanwhile, the motion information derived from the L1 reference picture may be the same as the L0 input temporal motion information of the current block. Accordingly, when searching for motion information to be used as the final L1 temporal motion information of the current block in the L1 reference picture, the encoder and the decoder may find only motion information that is not the same as the L0 input temporal motion information. For example, as in S1430, when the final L1 temporal motion information of the current block is derived based on the L1 reference picture, the encoder and the decoder only use motion information different from the L0 input temporal motion information of the current block. The final L1 temporal motion information can be used as In this case, the encoder and the decoder may select only motion information whose difference from the L0 input temporal motion information of the current block is less than or equal to a predetermined threshold as motion information to be used as the final L1 temporal motion information. Here, the predetermined threshold value may be determined based on mode information of the current block, motion information of the current block, mode information of the neighboring block, and/or motion information of the neighboring block, and may be determined in various ways.

In the above-described embodiment, the input temporal motion information input in step S1410 may include not only the input motion vector but also the input reference picture index. Here, the L0 input motion vector and the L1 input motion vector may be motion vectors temporally derived based on the collocated block as described above, and the L0 input reference picture index and the L1 input reference picture index are spatially derived from the reconstructed neighboring block. It may be a derived reference picture index. In this case, the L0 input reference picture index and the L1 input reference picture index may be set to the smallest non-negative value among the reference picture indexes of the reconstructed neighboring blocks. Meanwhile, as another example, the L0 input reference picture index and the L1 input reference picture index may be set to 0 regardless of the motion information of the reconstructed neighboring blocks.

When the L1 input motion vector is reset based on the L1 reference picture, a reset process may also be performed on the input reference picture index. As an example, the input reference picture index may be reset based on the reference picture index derived from the L1 reference picture as described above. As another example, L0 input temporal motion information (eg, L0 input motion vector, L0 input reference picture index, etc.) and L1 input temporal motion information (eg, L1 input motion vector, L1 input reference picture index, etc.) are the same In this case, the encoder and the decoder may reset both the L0 input reference picture index and the L1 input reference picture index to a value of 0 to use as final temporal motion information. This is because, when the L0 input temporal motion information and the L1 input temporal motion information are the same, there is a high probability that both the L0 reference picture index and the L1 reference picture index are 0.

Meanwhile, as described above, the encoder and the decoder may derive the final L1 temporal motion vector by resetting the L1 input motion vector value to the same value as the motion vector of the L1 reference picture. In this case, the motion vector of the L1 reference picture may be scaled and used according to the L1 input reference picture index and/or the reset L1 reference picture index. The L1 input reference picture index may be used as the final L1 reference picture index as it is without a reset process, or may be used as the final L1 reference picture index through the reset process as in the above-described embodiment. In this case, the reference picture corresponding to the motion vector of the L1 reference picture and the reference picture indicated by the final L1 reference picture index may be different from each other. In this case, the encoder and the decoder may perform scaling on the motion vector of the L1 reference picture and use the scaled motion vector as the final L1 temporal motion vector of the current block.

Also, in the above-described embodiments, the reset L1 temporal motion information may be the same as the L0 temporal motion information of the current block. Accordingly, when the reset L1 temporal motion information is the same as the L0 temporal motion information of the current block, the encoder and the decoder may reset the prediction direction information of the current block to unidirectional prediction. In this case, the encoder and the decoder may use only the L0 temporal motion information as the temporal motion information of the current block. This method can be applied to the present invention in combination with the above-described embodiments.

* FIG. 15 is a block diagram schematically illustrating an embodiment of an inter prediction apparatus capable of performing a process of deriving temporal motion information according to the embodiment of FIG. 14 . The inter prediction apparatus according to the embodiment of FIG. 15 may include a temporal motion information determiner 1510 , a motion predictor 1520 , and an L1 temporal motion information resetter 1530 .

As described above, the temporal motion information may be derived based on the motion information of the collocated block corresponding to the current block in the already reconstructed collocated picture. Here, the collocated picture may correspond to, for example, one picture among reference pictures included in the reference picture list. The encoder and the decoder may determine a predetermined relative position based on a block existing at the same spatial position as the current block in the collocated picture, and the determined relative position (eg, spatially identical to the current block). The collocated block may be derived based on the position inside and/or outside the block existing in the position). The temporal motion information derived based on the collocated block may include prediction direction information, L0 reference picture number, L1 reference picture number, L0 motion vector, L1 motion vector, and the like.

Referring to FIG. 15 , referring to FIG. 11 , the temporal motion information determining unit 1510 determines whether the L0 input temporal motion information and the L1 input temporal motion information are the same in the input temporal motion information, that is, the L0 input reference picture number and L1. It may be determined whether the input reference picture numbers are the same and the L0 input motion vector and the L1 input motion vector are the same.

When the L0 input temporal motion information and the L1 input temporal motion information are not the same, the inter prediction apparatus may use the input temporal motion information as the temporal motion information of the current block as it is. When AMVP is applied, the temporal motion information of the current block may be determined or registered as a prediction motion vector candidate for the current block. Also, when the merge is applied, the temporal motion information of the current block may be determined or registered as a merge candidate for the current block. When the L0 input temporal motion information and the L1 input temporal motion information are the same, a process by the motion predictor 1520 may be performed.

Meanwhile, the temporal motion information determining unit 1510 determines whether the L0 input reference picture number and the L1 input reference picture number are identical or the prediction direction of the collocated block, not whether the L0 input temporal motion information and the L1 input temporal motion information are identical. You may. For example, if the L0 input reference picture number and the L1 input reference picture number are not the same, the input temporal motion information may be used as the temporal motion information of the current block as it is, and the L0 input reference picture number and the L1 input reference picture number are In the same case, the process by the motion predictor 1520 may be performed. As another example, when the prediction direction of the collocated block is bidirectional prediction, the input temporal motion information may be used as the temporal motion information of the current block as it is, and when the prediction direction of the collocated block is unidirectional prediction, the process by the motion prediction unit 1520 may be performed.

Referring back to FIG. 15 , when the L0 input temporal motion information and the L1 input temporal motion information are the same, the motion predictor 1520 may search for or derive motion information to be used as the final L1 temporal motion information in the L1 reference picture.

As an embodiment, the motion predictor 1520 may derive a region that best matches or is similar to an input block (original block) corresponding to the current block in the L1 reference picture. In this case, the motion predictor 1520 may derive motion information to be used as the final L1 temporal motion information of the current block based on the derived region. For example, the motion predictor 1520 may derive a motion vector to be used as the final L1 temporal motion information based on the position of the block corresponding to the derived region and the position of the current block, and the motion vector corresponding to the derived region may be A reference picture index (and/or reference picture number) to be used as the final L1 temporal motion information may be derived based on the block.

Meanwhile, when the motion predictor 1520 corresponds to a component of the encoder, the encoder may include the motion information derived from the L1 reference picture by the above method in the bitstream and transmit it to the decoder. The motion information derived from the L1 reference picture may include a reference picture index and a motion vector, and the encoder may independently transmit the reference picture index and the motion vector to the decoder. In this case, the encoder may obtain a difference value between the actual motion vector of the current block and the motion vector derived from the L1 reference picture, and then transmit the difference value to the decoder.

In this case, the decoder may determine motion information to be used as the final L1 temporal motion information based on the transmitted motion information. That is, when the motion prediction unit 1520 corresponds to a component on the decoder side, the motion prediction unit 1520 determines the final result based on the motion information input from the outside (eg, motion information transmitted from the encoder). Motion information to be used as the L1 temporal motion information may be determined.

Referring back to FIG. 15 , the L1 temporal motion information resetter 1530 may use the motion information derived from the L1 reference picture as the final L1 temporal motion information of the current block. That is, at this time, the L1 temporal motion information resetting unit 1530 may reset the L1 input temporal motion information value to the value of the motion information (motion information derived from the L1 reference picture) derived by the motion prediction unit 1520 . .

The above-described embodiments of FIGS. 6 to 15 may be applied individually, or may be applied in combination in various ways according to the encoding mode of each block. Hereinafter, in embodiments to be described below, a block in which the encoding mode is a merge mode is referred to as a merge block. A block other than the merge block may include, for example, a block in which the encoding mode is the AMVP mode. Also, in embodiments to be described later, the current block may correspond to one of a merge block or a non-merged block in some cases.

As an embodiment, the method for deriving temporal motion information according to the embodiments of FIGS. 6 to 8 may be applied to the merge block, and the method for deriving the motion information according to FIGS. 12 and 13 may be applied to blocks other than the merge block. In this case, when the L0 input temporal motion information and the L1 input temporal motion information are the same in the merge block, the motion information of the reconstructed neighboring block may be used as the final L1 temporal motion information of the current block. Also, in a block other than the merge block, L0 motion information of the current block and L1 motion information of the current block may be independently set. For example, one of the L0 motion information of the current block and the L1 motion information of the current block may be spatially derived based on the reconstructed neighboring block, and the other may be derived temporally based on the collocated block.

In another embodiment, the method for deriving temporal motion information according to the embodiment of FIGS. 9 to 11 may be applied to the merge block, and the method for deriving the temporal motion information according to FIGS. 14 and 15 may be applied to a block other than the merge block. At this time, in the merge block, when the L0 input temporal motion information and the L1 input temporal motion information are the same, the motion information of the newly derived collocated block, not the collocated block used to derive the input temporal motion information, is the final L1 temporal motion information of the current block. can be used as Also, in a block other than the merge block, when the L0 input temporal motion information and the L1 input temporal motion information are the same, the motion information of the L1 reference picture may be used as the final L1 temporal motion information of the current block.

As another embodiment, the method for deriving temporal motion information according to the embodiments of FIGS. 6 to 8 may be applied to the merge block, and the method for deriving the temporal motion information according to FIGS. 14 and 15 may be applied to blocks other than the merge block. In this case, when the L0 input temporal motion information and the L1 input temporal motion information are the same in the merge block, the motion information of the reconstructed neighboring block may be used as the final L1 temporal motion information of the current block. Also, in a block other than the merge block, when the L0 input temporal motion information and the L1 input temporal motion information are the same, the motion information of the L1 reference picture may be used as the final L1 temporal motion information of the current block.

The combination of the embodiments of FIGS. 6 to 15 is not limited to the above-described embodiment, and various types of combinations may be provided as well as the above-described embodiments according to implementation and/or necessity.

In the above-described 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 the steps, and some steps may occur in a different order or at the same time as other steps as described above. can In addition, those of ordinary skill in the art will recognize that the steps shown in the flowchart are not exclusive, other steps may be included, or that one or more steps in the flowchart may be deleted without affecting the scope of the present invention. You will understand.

The above-described embodiments include examples of various aspects. It is not possible to describe every possible combination for representing the various aspects, but one of ordinary skill 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 (3)

In the video decoding method,
determining a prediction mode of the current block as an inter prediction mode;
deriving a spatial motion vector of the current block by using the spatially neighboring blocks of the current block;
determining a collocated picture of the current block based on collocated picture information; Here, the collocated picture information is information specifying a picture determined as the collocated picture of the current block from among the reference pictures in the reference picture list,
deriving a temporal motion vector based on a motion vector of a collocated block corresponding to the current block in the collocated picture; Here, the collocated block is determined according to either a first block including the position of the lower right-bottom pixel of the current block or a second block including the center position of the current block,
generating a predicted motion vector candidate list for the current block using the spatial motion vector and the temporal motion vector;
deriving a motion vector of the current block based on the predicted motion vector candidate list; and
obtaining a prediction sample for the current block based on the derived motion vector;
obtaining a residual sample for the current block; and
Comprising the step of obtaining a reconstructed sample for the current block by using the prediction sample and the residual sample for the current block,
The video decoding method, characterized in that the collocated picture information is transmitted from an encoder through a bitstream. In the video encoding method,
determining a prediction mode of the current block as an inter prediction mode;
deriving a motion vector of the current block;
obtaining a prediction sample corresponding to the current block by performing motion compensation on the current block based on the derived motion vector;
obtaining a residual sample for the current block based on the prediction sample;
deriving a spatial motion vector of the current block by using the spatially neighboring blocks of the current block;
determining a collocated picture of the current block;
deriving a temporal motion vector based on a motion vector of a collocated block corresponding to the current block in the collocated picture; Here, the collocated block is determined according to either a first block including the position of the lower right-bottom pixel of the current block or a second block including the center position of the current block,
generating a predicted motion vector candidate list for the current block using the spatial motion vector and the temporal motion vector;
obtaining a reconstructed sample for the current block using the prediction sample and the residual sample; and
Coding the collocated picture information for the collocated picture into a bitstream,
The collocated picture information is information specifying a picture determined as the collocated picture of the current block from among reference pictures in a reference picture list, and is transmitted to a decoder through the bitstream. A recording medium comprising a bitstream decoded by an image decoding method, the recording medium comprising:
The bitstream includes collocated picture information for a collocated picture,
determining a prediction mode of the current block as an inter prediction mode based on information included in the bitstream;
deriving a motion vector of the current block;
obtaining a prediction sample corresponding to the current block by performing motion compensation on the current block based on the derived motion vector;
obtaining a residual sample for the current block based on the prediction sample;
deriving a spatial motion vector of the current block by using the spatially neighboring blocks of the current block;
determining the collocated picture of the current block;
deriving a temporal motion vector based on a motion vector of a collocated block corresponding to the current block in the collocated picture; Here, the collocated block is determined according to either a first block including the position of the lower right-bottom pixel of the current block or a second block including the center position of the current block,
generating a predicted motion vector candidate list for the current block using the spatial motion vector and the temporal motion vector;
obtaining a reconstructed sample for the current block using the prediction sample and the residual sample; and
encoding the collocated picture information for the collocated picture into the bitstream;
The collocated picture information is information specifying a picture determined as a collocated picture of the current block among reference pictures in a reference picture list, and is signaled to a decoder through the bitstream. Bit generated by the video encoding method A recording medium for storing streams.

Images