CN113455000B - Bidirectional prediction method and video decoding equipment - Google Patents

Abstract

A bi-directional inter prediction method and an image decoding apparatus are disclosed. According to an embodiment of the present invention, there is provided a bi-prediction method for inter-predicting a current block using any one of a plurality of bi-prediction modes, including: decoding mode information from a bitstream, which indicates whether to apply a first mode included in a plurality of bi-prediction modes to a current block; when the mode information indicates that the first mode is applied to the current block, decoding first motion information including differential motion vector information and prediction motion vector information and second motion information excluding at least a portion of the differential motion vector information and the prediction motion vector information from the bitstream; deriving a first motion vector based on the first motion information, deriving a second motion vector based on at least a portion of the first motion information and the second motion information; and predicting the current block by using the reference block indicated by the first motion vector in the first reference picture and the reference block indicated by the second motion vector in the second reference picture.

Description

Bidirectional prediction method and video decoding equipment

Technical field

The present invention relates to video encoding and decoding, and more specifically, to a bidirectional prediction method and a video decoding device that improve encoding and decoding efficiency by effectively expressing motion information.

Background technique

Since the volume of video data is larger than that of voice data or still image data, storing or transmitting video data without compression requires a large amount of hardware resources, including memory.

Therefore, when storing or transmitting video data, an encoder is usually used to compress the video data for storage or transmission. The decoder then receives the compressed video data and decompresses and reproduces the video data. Compression technologies used for such videos include H.264/AVC and High Efficiency Video Coding (HEVC), which are about 40% more efficient than H.264/AVC.

However, the size, resolution and frame rate of videos are gradually increasing, and accordingly the amount of data to be encoded is also increasing. Therefore, new compression technologies with better coding efficiency and higher image quality than existing compression technologies are needed.

Contents of the invention

technical problem

It is an object of the present invention to provide improved video encoding and decoding techniques, and more specifically, to provide techniques that improve encoding and decoding efficiency by using motion information in a specific direction to infer motion information in other directions.

Technical solutions

According to at least one aspect, the present disclosure provides a method of inter-predicting a current block using any one of a plurality of bi-directional prediction modes. The method includes decoding mode information from a bitstream, the mode information indicating whether to apply a first mode included in a plurality of bidirectional prediction modes to a current block. When the mode information indicates that the first mode is applied to the current block, the method further includes decoding, from the bitstream, first motion information including differential motion vector information and predicted motion vector information regarding the first motion vector and excluding information regarding the second motion. differential motion vector information of the vector and second motion information that predicts at least a portion of the motion vector information; and deriving a first motion vector based on the first motion information, and deriving a second motion vector based on at least a portion of the first motion information and based on the second motion information. Motion vector. The method also includes predicting the current block using a reference block in the first reference picture indicated by the first motion vector and a reference block in the second reference picture indicated by the second motion vector.

According to another aspect, the present disclosure provides a video decoding device. The device includes a decoder configured to decode mode information from a bitstream, the mode information indicating whether a first mode included in a plurality of bidirectional prediction modes is applied to a current block. When the mode information indicates that the first mode is applied to the current block, the decoder decodes from the bitstream first motion information including differential motion vector information and predicted motion vector information about a first motion vector and second motion information excluding at least a portion of differential motion vector information and predicted motion vector information about a second motion vector. The device includes a prediction unit configured to derive a first motion vector based on the first motion information, and derive a second motion vector based on at least a portion of the first motion information and the second motion information. The predictor is configured to predict the current block using a reference block indicated by the first motion vector in a first reference picture and a reference block indicated by the second motion vector in a second reference picture.

Technical effect

As described above, according to embodiments of the present invention, the bit efficiency of motion representation can be improved by using motion in a specific direction to infer motion in other directions.

Description of drawings

1 is an exemplary block diagram of a video encoding device capable of implementing the techniques of the present disclosure.

Figure 2 exemplarily shows the block partition structure using the QTBTTT structure.

Figure 3 exemplarily shows various intra prediction modes.

4 is an exemplary block diagram of a video decoding device capable of implementing the techniques of the present disclosure.

FIG. 5 is a diagram for describing bidirectional prediction according to an embodiment of the present invention.

6 is a diagram for describing deriving motion using a symmetrical relationship between differential motion vectors according to an embodiment of the present invention.

7 and 8 are diagrams for describing the use of linear relationships to derive motion according to an embodiment of the present invention.

9 to 18 are diagrams for describing derivation motion according to various embodiments of the present invention.

19 and 20 are flowcharts for describing derivation of motion using reference pictures determined at a high level according to an embodiment of the present invention.

Detailed ways

Hereinafter, some embodiments of the present disclosure will be described in detail with reference to the accompanying drawings. It should be noted that when reference numbers are added to constituent elements in various drawings, similar reference numbers refer to similar elements although the elements are shown in different drawings. Furthermore, in the following description of the present disclosure, detailed descriptions of known functions and configurations incorporated herein will be omitted to avoid obscuring the subject matter of the present disclosure.

FIG1 is an exemplary block diagram of a video encoding device capable of implementing the technology of the present disclosure. Hereinafter, a video encoding device and elements of the device will be described with reference to FIG1 .

The video encoding device includes a block partitioner 110, a predictor 120, a subtractor 130, a transformer 140, a quantizer 145, an encoder 150, an inverse quantizer 160, an inverse transformer 165, an adder 170, a filter unit 180, and a memory 190 .

Each element of the video encoding device may be implemented in hardware or software, or a combination of hardware and software. The functions of each component can be implemented by software, and a microprocessor can be implemented to execute the software functions corresponding to each component.

A video consists of multiple pictures. Each image is segmented into multiple regions, and encoding is performed on each region. For example, a picture is divided into one or more tiles and/or slices. Here, one or more tiles can be defined as a tile group. Each tile or slice is divided into one or more coding tree units (CTUs). Each CTU is divided into one or more coding units (CU) according to a tree structure. Information that applies to each CU is encoded as the syntax of the CU, and information that applies collectively to the CUs contained in a CTU is encoded as the syntax of the CTU. Furthermore, information that is commonly applied to all blocks in a tile is encoded into the syntax of the tile or is encoded into the syntax of a tile group that is a collection of multiple tiles, and is applied to all blocks that make up a picture. The information is encoded in the picture parameter set (PPS) or picture header. In addition, information commonly referenced by multiple pictures is encoded in a sequence parameter set (SPS). In addition, information commonly referenced by one or more SPS is encoded in a video parameter set (VPS).

Block partitioner 110 determines the size of the coding tree unit (CTU). Information about the size of the CTU (CTU size) is encoded into the syntax of SPS or PPS and sent to the video decoding device.

The block divider 110 divides each picture constituting the video into a plurality of CTUs having a predetermined size, and then recursively divides the CTUs using a tree structure. In the tree structure, leaf nodes serve as coding units (CUs), which are the basic units of coding.

The tree structure can be a quadtree (QT) where a node (or parent node) is split into four child nodes (or child nodes) of the same size, a binary tree (BT) where a node is split into two child nodes, where a node A ternary tree (TT) divided into three child nodes in a ratio of 1:2:1, or a structure formed by a combination of two or more of the QT structure, BT structure, and TT structure. For example, a QTBT (quadtree plus binary tree) structure or a QTBTTT (quadtree plus binary tree ternary tree) structure may be used. Here, BTTT can be collectively referred to as multi-type tree (MTT).

Figure 2 shows the QTBTTT split tree structure. As shown in Figure 2, the CTU can initially be segmented in a QT structure. QT splitting can be repeated until the split block size reaches the minimum block size (MinQTSize) of leaf nodes allowed in QT. A first flag (QT_split_flag) indicating whether each node of the QT structure is split into four nodes of the lower layer is encoded by the encoder 150 and signaled to the video decoding device. When the leaf node of QT is not larger than the maximum block size (MaxBTSize) of the root node allowed in BT, it can be further divided in one or more of the BT structure or the TT structure. In BT structure and/or TT structure, there can be multiple splitting directions. For example, there may be two directions as horizontal splitting and vertical splitting of node blocks. As shown in FIG. 2 , when MTT splitting starts, the encoder 150 sets a second flag (mtt_split_flag) indicating whether a node is split, a flag indicating a splitting direction (vertical or horizontal), and/or a splitting type (binary or triple) mark to encode and signal the video decoding device.

As another example of a tree structure, when a block is split using the QTBTTT structure, a CU split flag (split_cu_flag) indicating that the block has been split and a QT split flag (split_qt_flag) indicating that the split type is QT split are set by the encoder 150 The information is encoded and signaled to the video decoding device. When the value of split_cu_flag indicates that the block has not been split, the block of the node becomes a leaf node in the split tree structure and is used as a coding unit (CU) which is a basic unit of encoding. When the value of split_cu_flag indicates that the block has not been split, the value of split_qt_flag is used to distinguish whether the split type is QT or MTT. When the split type is QT, there is no additional information. When the split type is MTT, a flag (mtt_split_cu_vertical_flag) indicating the MTT split direction (vertical or horizontal) and/or a flag (mtt_split_cu_binary_flag) indicating the MTT split type (binary or trifurcated) is encoded by the encoder 150 and signaled to the video Decoding equipment.

As another example of a tree structure, when using QTBT, there can be two splitting types, which are block horizontal splitting (i.e. symmetrical horizontal splitting) and vertical splitting (i.e. symmetrical vertical splitting) of nodes into the same size of two blocks. A split flag (split_flag) indicating whether each node of the BT structure is split into lower-layer blocks and split type information indicating the split type are encoded by the encoder 150 and sent to the video decoding device. There can be additional types that split a block of nodes into two asymmetric blocks. The asymmetric division type may include a type in which a block is divided into two rectangular blocks with a size ratio of 1:3, and a type in which a node block is divided diagonally.

CUs can be of various sizes depending on the QTBT or QTBTTT partitioning of the CTU. Hereinafter, the block corresponding to the CU to be encoded or decoded (ie, the leaf node of QTBTTT) is referred to as the "current block".

The predictor 120 predicts the current block to generate a prediction block. Predictors 120 include intra predictors 122 and inter predictors 124 .

In general, each current block in the picture can be predictively encoded. Prediction of the current block may be performed using intra prediction techniques (performed based on data from the picture containing the current block) or inter prediction techniques (performed based on data from pictures encoded before the picture containing the current block). Inter prediction includes both unidirectional prediction and bidirectional prediction.

The intra predictor 122 predicts pixels in the current block using pixels (reference pixels) located around the current block in the current picture including the current block. There are multiple intra prediction modes depending on the prediction direction. For example, as shown in FIG. 3 , a plurality of intra prediction modes may include a non-directional mode including a planar mode and a DC mode and 65 directional modes. For each prediction mode, the neighboring pixels and formulas to be used are defined differently.

Intra predictor 122 may determine the intra prediction mode to use when encoding the current block. In some examples, intra predictor 122 may encode the current block using several intra prediction modes and select an appropriate intra prediction mode to use from among the tested modes. For example, the intra predictor 122 may calculate a rate-distortion value using rate-distortion analysis of several tested intra-prediction modes, and may select an intra-prediction mode with the best rate-distortion characteristics among the tested modes.

The intra predictor 122 selects an intra prediction mode from among a plurality of intra prediction modes and predicts the current block using neighboring pixels (reference pixels) determined according to the selected intra prediction mode and a formula. The information about the selected intra prediction mode is encoded by the encoder 150 and sent to the video decoding device.

The inter-frame predictor 124 generates a prediction block of the current block through motion compensation processing. The inter-frame predictor searches for a block most similar to the current block in a reference picture that is encoded and decoded earlier than the current picture, and generates a prediction block of the current block based on the searched block. Then, the inter-frame predictor generates a motion vector corresponding to the displacement between the current block in the current picture and the prediction block in the reference picture. Generally, motion estimation is performed on the luminance component, and for both the luminance component and the chrominance component, a motion vector calculated based on the luminance component is used. Motion information including information about the reference picture used to predict the current block and information about the motion vector is encoded by the encoder 150 and sent to the video decoding device.

Subtractor 130 generates a residual block by subtracting the prediction block generated by intra predictor 122 or inter predictor 124 from the current block.

The transformer 140 transforms the residual signal in the residual block having pixel values in the spatial domain into transform coefficients in the frequency domain. The transformer 140 may transform the residual signal in the residual block using the total size of the current block as a transform unit. Alternatively, the transformer may divide the residual block into sub-blocks of the transform area and the non-transform area, and transform the residual signal using only the sub-blocks of the transform area as transform units. Here, the transformation area sub-block may be one of two rectangular blocks with a size ratio based on the horizontal axis (or vertical axis) of 1:1. In this case, a flag (cu_sbt_flag), direction (vertical/horizontal) information (cu_sbt_horizontal_flag), and/or position information (cu_sbt_pos_flag) indicating that only the sub-block has been transformed is encoded and signaled to the video decoding device by the encoder 150 . In addition, the size of the transform region sub-block may have a size ratio of 1:3 based on the horizontal axis (or vertical axis). In this case, a flag (cu_sbt_quad_flag) for distinguishing the divisions is additionally encoded and signaled to the video decoding device by the encoder 150 .

The quantizer 145 quantizes the transform coefficient output from the transformer 140 and outputs the quantized transform coefficient to the encoder 150 .

The encoder 150 generates a bitstream by encoding the quantized transform coefficients using a coding method such as context-based adaptive binary arithmetic coding (CABAC). The encoder 150 encodes information related to block partitioning, such as CTU size, CU partition flag, QT partition flag, MTT partition direction, and MTT partition type, so that the video decoding device and the video encoding device partition the blocks in the same manner.

In addition, the encoder 150 encodes information on a prediction type indicating whether the current block is encoded by intra prediction or inter prediction, and the intra prediction information (ie, information on an intra prediction mode) according to the prediction type Or inter prediction information (information about reference pictures and motion vectors) is encoded.

The inverse quantizer 160 inversely quantizes the quantized transform coefficients output from the quantizer 145 to generate transform coefficients. The inverse transformer 165 transforms the transform coefficient output from the inverse quantizer 160 from the frequency domain to the spatial domain and reconstructs the residual block.

The adder 170 adds the reconstructed residual block and the prediction block generated by the predictor 120 to reconstruct the current block. The reconstructed pixels in the current block are used as reference pixels for intra prediction of the next block.

Filter unit 180 filters the reconstructed pixels to reduce blocking artifacts, ringing artifacts, and blurring artifacts resulting from block-based prediction and transform/quantization. Filter unit 180 may include a deblocking filter 182 and a sample adaptive offset (SAO) filter 184.

The deblocking filter 180 filters the boundaries between reconstructed blocks to remove blocking artifacts caused by block-by-block encoding/decoding, and the SAO filter 184 additionally filters the deblocking filtered video. The SAO filter 184 is a filter for compensating for differences between reconstructed pixels and original pixels caused by lossy encoding.

The reconstructed blocks filtered by the deblocking filter 182 and the SAO filter 184 are stored in the memory 190 . Once all blocks in a picture have been reconstructed, the reconstructed picture is used as a reference picture for inter prediction of the next picture to be encoded.

4 is an exemplary functional block diagram of a video decoding device capable of implementing the techniques of the present disclosure. In the following, a video decoding device and elements of the device will be described with reference to FIG. 4 .

The video decoding device may include a decoder 410, an inverse quantizer 420, an inverse transformer 430, a predictor 440, an adder 450, a filter unit 460, and a memory 470.

Similar to the video encoding device of FIG. 1, each element of the video decoding device may be implemented as hardware or software, or may be implemented as a combination of hardware and software. In addition, the function of each element may be implemented as software, and a microprocessor may be implemented to execute the function of the software corresponding to each element.

The decoder 410 determines the current block to be decoded by decoding the bit stream received from the video encoding device and extracting information related to block partitioning, and extracts prediction information required to reconstruct the current block and information on the residual signal .

The decoder 410 extracts information about the CTU size from the sequence parameter set (SPS) or picture parameter set (PPS), determines the size of the CTU, and segments the picture into CTUs of the determined size. Then, the decoder determines the CTU as the uppermost layer (ie, the root node of the tree structure), and extracts segmentation information about the CTU to segment the CTU using the tree structure.

For example, when using the QTBTTT structure to split a CTU, first extract the first flag (QT_split_flag) related to QT splitting, and split each node into four nodes in the lower layer. Then, for the node corresponding to the leaf node of the QT, extract the second flag (MTT_split_flag) related to the MTT split and the information about the split direction (vertical/horizontal) and/or the split type (binary/trifurcated), and use The MTT structure divides leaf nodes. In this way, each node below the leaf node of QT is recursively split in a BT or TT structure.

As another example, when a CTU is split using the QTBTTT structure, a CU split flag (split_cu_flag) indicating whether the CU is split is first extracted. If the corresponding block is split, the QT split flag (split_qt_flag) is extracted. When the split type is not QT but MTT, a flag (mtt_split_cu_vertical_flag) indicating the MTT split direction (vertical or horizontal) and/or a flag (mtt_split_cu_binary_flag) indicating the MTT split type (binary or trifurcated) is additionally extracted. During the splitting process, each node can undergo zero or more recursive QT splits and then zero or more recursive MTT splits. For example, a CTU can be split immediately by MTT, or it can be split multiple times by QT only.

As another example, when the CTU is split using the QTBT structure and the first flag (QT_split_flag) related to QT splitting, and each node is split into four nodes of the lower layer. For the node corresponding to the leaf node of QT, extract the split_flag and split direction information indicating whether the node is further BT split.

Once the current block to be decoded is determined through tree structure partitioning, the decoder 410 extracts information about the prediction type indicating whether the current block is subjected to intra prediction or inter prediction. When the prediction type information indicates intra prediction, the decoder 410 extracts syntax elements of intra prediction information (intra prediction mode) of the current block. When the prediction type information indicates inter prediction, the decoder 410 extracts syntax elements of inter prediction information, that is, information indicating a motion vector and a reference picture to which the motion vector refers.

The decoder 410 extracts information on the quantized transform coefficient of the current block as information on the residual signal.

The inverse quantizer 420 inversely quantizes the quantized transform coefficients, inversely transforms the inversely quantized transform coefficients from the frequency domain to the spatial domain, and reconstructs the residual signal to generate a residual block of the current block.

In addition, when the inverse transformer 430 inversely transforms only the local area (sub-block) of the transform block, extracts a flag (cu_sbt_flag) indicating that only the sub-block of the transform block has been transformed, and the direction information (vertical/ Horizontal) (cu_sbt_horizontal_flag) and/or sub-block position information (cu_sbt_pos_flag). Then, the residual signal is reconstructed by inversely transforming the transform coefficients of the sub-block from the frequency domain to the spatial domain. For areas that have not been inversely transformed, the residual signal is filled with "0". Thereby, the final residual block of the current block is created.

Predictors 440 may include intra predictors 442 and inter predictors 444. Intra predictor 442 is activated when the prediction type of the current block is intra prediction, and inter predictor 444 is activated when the prediction type of the current block is inter prediction.

The intra predictor 442 determines an intra prediction mode of the current block among a plurality of intra prediction modes based on the syntax elements of the intra prediction mode extracted from the decoder 410 and based on the reference pixels around the current block according to the intra prediction mode. Predict the current block.

The inter predictor 444 determines the motion vector of the current block and the reference picture to which the motion vector refers based on the syntax elements of the intra prediction mode extracted from the decoder 410, and predicts the current block based on the motion vector and the reference picture.

The adder 450 reconstructs the current block by adding the residual block output from the inverse transformer and the prediction block output from the inter predictor or the intra predictor. The reconstructed pixels in the current block are used as reference pixels for intra prediction of blocks to be decoded later.

Filter unit 460 may include a deblocking filter 462 and an SAO filter 464. Deblocking filter 462 performs deblocking filtering on boundaries between reconstructed blocks to remove blocking artifacts caused by block-by-block decoding. The SAO filter 464 performs additional filtering on the reconstructed blocks after deblocking filtering to compensate for differences between the reconstructed pixels and original pixels caused by lossy encoding. The reconstructed blocks filtered by deblocking filter 462 and SAO filter 464 are stored in memory 470 . When all blocks in a picture are reconstructed, the reconstructed picture is used as a reference picture for inter prediction of blocks in the picture to be encoded later.

The inter-picture prediction encoding/decoding method (inter prediction method) of the HEVC standard can be classified into a skip mode, a merge mode, and an adaptive (or advanced) motion vector predictor (AMVP) mode.

In skip mode, an index value indicating one of the motion information candidates of the adjacent block is signaled. In the merge mode, an index value indicating one of the motion information candidates of the adjacent block and information obtained by encoding the prediction residual are signaled. In AMVP mode, the motion information of the current block and the information obtained by encoding the residual after prediction are signaled. The motion information signaled in AMVP mode includes the motion information of neighboring blocks (motion vector predictor (mvp)) and the difference between the motion information (mvp) and the motion information (mv) of the current block (motion vector difference). (mvd)).

Describing the motion information signaled in the AMVP mode in more detail, the motion information may include reference picture information (reference picture index), predicted motion vector (mvp) information, and differential motion vector (mvd) information. In the case of bidirectional prediction, the above information is signaled separately for each direction. Table 1 below shows the syntax elements for reference picture information, mvp information, and mvd information signaled for each direction.

[Table 1]

In Table 1 above, inter_pred_idc is a syntax element indicating the prediction direction (prediction direction information), and can indicate uni-L0 (uni-L0), uni-L1 (uni-L1), and bi-prediction (bi-prediction) any of them. According to the present invention, since motion information in a specific direction is derived from motion information in another direction, inter_pred_idc indicates bidirectional prediction. ref_idx_10 is a syntax element (reference picture information) indicating a reference picture in the L0 direction, and a reference picture used for predicting the current block among reference pictures included in the reference picture list 0 is specified by this syntax element. ref_idx_l1 is a syntax element (reference picture information) indicating a reference picture in the L1 direction, and the reference picture used for predicting the current block among the reference pictures included in the reference picture list 1 is specified by this syntax element. mvp_l0_flag is a syntax element (mvp information) indicating an mvp for the L0 direction, and the mvp to be used for prediction in the L0 direction of the current block is specified by this syntax element. mvp_l1_flag is a syntax element (mvp information) indicating an mvp for the L1 direction, and the mvp to be used for prediction in the L1 direction of the current block is specified by this syntax element.

The syntax elements that make up the mvd information are shown in Table 2 below.

[Table 2]

In Table 2 above, abs_mvd_greater0_flag is a syntax element indicating whether the absolute value (size) of mvd exceeds 0, and abs_mvd_greater1_flag is a syntax element indicating whether the absolute value (size) of mvd exceeds 1. In addition, abs_mvd_minus2 is a syntax element indicating a value obtained by subtracting 2 from the absolute value of mvd, and mvd_sign_flag corresponds to a syntax element indicating a sign of mvd.

As shown in Table 2, mvd is represented by a syntax element (abs_mvd_greater0_flag, abs_mvd_greater1_flag, abs_mvd_minus2) indicating an absolute value of each of an x component and a y component and a syntax element (mvd_sign_flag) indicating a sign.

Table 3 below summarizes the information signaled from the video encoding device to the video decoding device for bidirectional prediction of the conventional AMVP mode based on what is described in Tables 1 and 2.

[table 3]

As shown in Table 3 above, in the traditional AMVP mode, in order to perform bidirectional prediction for the current block, reference picture information, mvp information, mvd information, etc. are signaled separately for each direction, which can be inefficient in terms of bit efficiency. .

The present invention relates to improving the bit efficiency of bidirectional prediction by inferring a reference picture used to predict a current block or using correlation between motion information in each direction to infer motion information in other directions from motion information in a specific direction. .

"Specific direction" indicates a direction in which motion information is inferred or derived based on information signaled from the video encoding device, and "other direction" indicates a direction in which motion information is inferred or derived based on motion information in a specific direction. In inferring motion information in other directions, at least some of the motion information in a specific direction and/or the information signaled from the video encoding device may be used. In this specification, it is described that the specific direction corresponds to the direction L0 and the other directions correspond to the direction L1, but the specific direction may correspond to any one of the directions L0 and L1, and the other directions may correspond to neither of the two directions. The rest of the directions in a specific direction. Hereinafter, a specific direction is called a first direction, and other directions are called a second direction. In addition, the motion vector in the first direction is called a first motion vector, and the motion vector in the second direction is called a second motion vector.

The correlation between multiple pieces of motion information may include a symmetrical relationship, a linear relationship, a proportional relationship, a picture order count (POC) difference relationship between reference pictures based on the current picture, etc. established between the multiple pieces of motion information. This correlation may be established for all multiple pieces of motion information, and may be established separately for each element included in the motion information (at least one of the reference picture information, the MVP information, and the MVD information). For example, a symmetrical relationship may be established between multiple pieces of MVD information in two directions, and a linear relationship may be established between the MVP information in two directions (indicated by the MVP_FLAG) and the MVD information in two directions. Here, establishing a linear relationship between the MVP information and the MVD information in two directions may be understood as establishing a linear relationship between motion vectors (motions) in two directions.

In conjunction with the names of the motion information referred to in this specification, the motion information in a specific direction (first direction) is referred to as the first motion information, and the motion information in the other direction (second direction) is referred to as the second motion information or the third motion information depending on the number or type of elements included. The third motion information is the motion information in the second direction, and may be motion information including the mvd information in the second direction and the mvp information in the second direction. The second motion information and the third motion information both correspond to the motion information in the second direction, but may be classified according to whether both the mvd information and the mvp information in the second direction are included or at least one of the mvp information and the mvd information is not included.

An embodiment of the invention for inferring motion in the second direction is illustrated in Figure 5 .

The video encoding device may signal the mode information (mode_info) by including the mode information (mode_info) in the bitstream. The bidirectional prediction modes proposed by the present invention may include a first mode that derives second motion information (motion_info_l1) from first motion information (motion_info_l0), a second mode that uses signaled information to derive third motion information (motion_info_l2), etc. .

mode_info may correspond to information indicating any one of a plurality of prediction modes included in a plurality of bidirectional prediction modes. Depending on the number of available bidirectional prediction modes, the mode information can be implemented in various forms such as flags or indexes. Hereinafter, description will be made on the premise that mode_info indicates a prediction mode used for bidirectional prediction of the current block among the first mode and the second mode. Under this premise, mode_info may correspond to information indicating whether the first mode is applied to the current block. In addition, the case where mode_info does not indicate application of the first mode may be the same as indicating not applying the first mode or indicating application of the second mode.

When mode_info indicates that the first mode is applied, the video encoding device may signal motion_info_10 and motion_info_11 by including motion_info_10 and motion_info_11 in the bitstream. Motion_info_10 may include differential motion vector information (mvd_10) in the first direction and predicted motion vector information (mvp_10_flag) in the first direction. Motion_info_11 may include some of mvd_11 and mvp_11_flag (in other words, motion_info_11 may not include at least some of mvd_11 and mvp_11_flag). On the other hand, when mode_info does not indicate that the first mode is applied (when mode_info indicates that the second mode is applied), the video encoding device may signal motion_info_10 and motion_info_12 by including motion_info_10 and motion_info_12 in the bitstream. Motion_info_12 may include both mvd_11 and mvp_11_flag.

The video decoding device (decoding unit) may decode mode_info from the bit stream (S530). When mode_info indicates that the first mode is applied (S540), since motion_info_l1 is included in the bitstream, the video decoding device can decode motion_info_l0 and motion_info_l1 from the bitstream (S550).

The video decoding device (prediction unit) may derive the first motion vector mv_l0 based on motion_info_l0, and derive the second motion vector mv_l1 based on motion_info_l0 and at least a part of motion_info_l1 (S560). Since motion_info_l0 includes mvd_l0 and mvp_l0_flag, mv_l0 can be derived by summing mvd_l0 and mvp_l0 in the following formula 1.

[Formula 1]

In Equation 1 above, mvx ₀ represents the x component of mv_l0, and mvy ₀ represents the y component of mv_l0. mvpx ₀ represents the x component of mvp_l0, and mvpy ₀ represents the y component of mvp_l0. mvdx ₀ represents the x component of mvd_l0, and mvdy ₀ represents the y component of mvd_l0.

Since motion_info_l1 does not include at least part of mvd_l1 and mvp_l1_flag, mv_l1 can be derived based on the correlation of motion. The detailed method of deriving mv_l1 will be described below.

The video decoding apparatus may use a first reference block indicated by mv_10 within a first reference picture (ref_10) that is a reference picture in the first direction, and within a second reference picture that is a reference picture in the second direction. The second reference block (ref_l1) indicated by mv_l1, thereby predicting the current block (generating the prediction block of the current block) (S570). ref_l0 and ref_l1 may be specified based on reference picture information (ref_idx_l0 and ref_idx_l1) signaled from the video encoding device, or may be derived based on the POC difference between the reference picture included in the reference picture list and the current picture. Specific implementations thereof will be described below.

Meanwhile, when mode_info does not indicate application of the first mode (when mode_info indicates application of the second mode) in operation S540, since motion_info_12 is included in the bitstream, the video decoding device may decode motion_info_10 and motion_info_12 from the bitstream (S590). In this case, the video decoding device may derive mv_l0 based on motion_info_l0 and mv_l1 based on motion_info_l2 (S560). In addition, the video decoding device may predict the current block by using the first reference block indicated by mv_l0 and the second reference block indicated by mv_l1 (S570).

According to an embodiment, the video encoding device may signal the enabling information (enabled_flag) by further including the enabling information (enabled_flag) in the bitstream. The enabled_flag may correspond to information indicating whether the first mode is enabled. When the enabled_flag indicates that the first mode is enabled, the video encoding device may encode the enabled_flag as a high-level syntax such as a sequence level, a picture level, a tile group level, and a slice level, and signal the mode_info of each prediction unit (block) by including the mode_info of each prediction unit (block) in the bitstream. In this way, whether the embodiment proposed by the present invention is applied can be set for each block.

When enabled_flag is encoded into the high-level syntax and mode_info is encoded in block units, the video decoding device may decode enabled_flag from the high-level syntax (S510), and when enabled_flag indicates that the first mode is enabled (S520), decode motion_info from the bitstream (S530). Meanwhile, when enabled_flag indicates that the first mode is not enabled, mode_info may not be decoded. In this case, the video decoding device may not apply the first mode to the current block by setting or estimating mode_info to "0" or "off" to indicate not to apply the first mode (S580).

In the following, the method proposed by the present invention will be described according to whether some of the reference picture information (ref_idx_l0 and ref_idx_l1), predicted motion vector information (mvp_l0_flag and mvp_l1_flag), differential motion vector information (mvd_l0 and mvd_l0 and mvd_l1) are included in the motion information. Various implementations.

In embodiments described below, motion_info_l0 may include mvd_l0 and mvp_l0_flag, and motion_info_l1 may not include at least some of mvd_l1 and mvp_l1_flag. In other words, motion_info_l0 may not include ref_idx_l0, and motion_info_l1 may not include one or more of ref_idx_l1, mvd_l1, and mvp_l1_flag.

First embodiment

The first embodiment corresponds to a method of inferring motion information by inferring mvd_l1 when ref_idx_l0, mvd_l0 and mvp_l0 are all included in motion_info_l0 and ref_idx_l1 and mvp_l1 are included in motion_info_l1.

In the first embodiment, the unsignaled mvd_l1 may be derived from mvd_l0. mvd_l1 may be derived based on the symmetric relationship established between mvd_l1 and mvd_l0. That is, mvd_l1 may be set or derived to a value symmetric to mvd_l0 (mvd_l1=-mvd_l0), and mv_l1 may be derived using the derived mvd_l1 and the signaled mvp_l1 (Formula 2).

[Formula 2]

(mvx ₁ , mvy ₁ )=(mvpx ₁ -mvdx ₀ , mvpy ₁ -mvdy ₀ )

The video encoding device may signal motion_info_l0 and motion_info_l1 (except mvd_l1 ) by including motion_info_l0 and motion_info_l1 (except mvd_l1 ) in the bitstream by the same process as described above. As shown in FIG. 6, the video decoding device can derive mv_l0 by using mvd_l0 and mvp_l0 included in motion_info_l0. Furthermore, the video decoding device can derive mv_l1 by using mvd_l1 (-mvd_l0) derived from mvd_l0 and mvp_l1 included in motion_info_l1.

The video decoding apparatus may predict a current block 620 located in a current picture 610 using a first reference block 630 indicated by mv_l0 within ref_l0 indicated by ref_idx_l0 and a second reference block 640 indicated by mv_l1 within ref_l1 indicated by ref_idx_l1.

Second embodiment

The second embodiment corresponds to a method of inferring motion information by inferring ref_l0 and ref_l1 when ref_idx_l0 is not included in motion_info_l0 and ref_idx_l1 is not included in motion_info_l1.

In the second embodiment, ref_10 and ref_11 may be determined or derived as the reference picture with the 0th index (located at the first position) among the reference pictures included in the reference picture list, or may be based on ref_l0 and ref_l1 are determined or derived from the POC difference between the reference picture and the current picture. Hereinafter, a method of deriving ref_l0 and ref_l1 based on the POC difference from the current picture will be described.

The video decoding device may select any of the reference pictures included in the reference picture list in the first direction based on a difference in POC value between the reference picture included in the reference picture list 0 (the reference picture list in the first direction) and the current picture. A reference picture and set the selected reference picture to ref_l0. For example, the video decoding device may set the reference picture with the smallest POC value difference from the current picture (the closest reference picture) as ref_10.

In addition, the video decoding device may select a reference picture included in the reference picture list in the second direction based on a difference in POC value between the reference picture included in the reference picture list 1 (the reference picture list in the second direction) and the current picture. Select any reference picture and set the selected reference picture to ref_l1. For example, the video decoding device may set the reference picture with the smallest POC value difference from the current picture (the closest reference picture) as ref_l1.

The video decoding device may sequentially or in parallel compare the POC values of the reference pictures included in the reference picture list with the POC value of the current picture to select any one reference picture. When selecting the closest reference picture by sequentially comparing reference pictures included in the reference picture list, the video decoding device may virtually set the index value of the reference picture to an index value not assigned to the reference picture list (for example, -1 ) and then compare the reference images sequentially.

The reference picture selected from the reference picture list in the first direction and the reference picture selected from the reference picture list in the second direction may have a forward POC value or a backward POC value relative to the POC value of the current picture. That is, the reference picture selected from the reference picture list in the first direction and the reference picture selected from the reference picture list in the second direction may be composed of a pair of forward reference pictures and backward reference pictures.

When ref_l0 and ref_l1 are derived, the video decoding device may predict the current block using the first reference block 630 indicated by mv_l0 in ref_l0 and the second reference block 640 indicated by mv_l1 in ref_l1.

According to an embodiment, the process of determining ref_l0 and ref_l1 may be performed at a higher level than the level of the current block. That is, among the elements contained in motion_info_l0 and motion_info_l1, the remaining elements except ref_l0 and ref_l1 can be derived or determined in block units, and ref_l0 and ref_l1 can be determined in high-level units. Here, the high level may be a level higher than the block level, such as picture level, tile group level, slice level, tile level, and coding tree unit (CTU) level.

The second embodiment may be implemented in combination with the first embodiment described above or the embodiment to be described below. That is, although it has been described that ref_idx_l0 and ref_idx_l1 are signaled in the first embodiment, when the second embodiment is applied, ref_idx_l0 and ref_idx_l1 are not signaled in the first embodiment, so the video decoding device itself can Derive ref_l0 and ref_l1.

Third embodiment

The third embodiment corresponds to a method of inferring second motion information from first motion information based on a linear relationship established between motion in the first direction and motion in the second direction.

The video encoding device may signal the video decoding device motion_info_lO by including motion_info_l0 in the bitstream. motion_info_l0 may include mvp_l0_flag, mvd_l0 and/or ref_idx_l0. The information included in motion_info_10 may be different for each embodiment to be described later.

The video decoding device may decode motion_info_10 from the bit stream (S710). The video decoding device may infer or derive mv_l0 by using mvp_l0_flag and mvd_l0 (S720). mv_l0 can be derived by adding mvp_l0 and mvd_l0 as in Equation 1 described above. Here, mvp_l0 may correspond to the motion vector of the adjacent block indicated by the decoded mvp_l0_flag.

When mv_l0 is derived, the video decoding device may derive mv_l1 by using ref_l0, ref_l1, and mv_l0 (S730). The derived mv_l1 may correspond to a motion vector having a linear relationship with mv_l0. ref_10 may be a reference picture indicated by ref_idx_10 signaled from the video encoding device or a separately defined reference picture. In addition, ref_l1 may be a reference picture indicated by ref_idx_l1 signaled from the video encoding device or a separately defined reference picture.

mv_l1 can be derived by applying the proportional relationship between "the difference in the POC value between the current picture 610 and ref_l0" and "the difference in the POC value between the current picture 610 and ref_l1" to mv_l0 as shown in the following formula 3 .

[Formula 3]

In formula 3, mvx ₁ represents the x component of mv_l1, and mvy ₁ represents the y component of mv_l1. POC ₀ represents the POC value of ref_l0, POC ₁ represents the POC value of ref_l1, and POC _curr represents the POC value of the current picture 610 including the current block 620. In addition, POC _curr - POC ₀ represents the difference in POC values between ref_10 and the current picture 610 , and POC _curr - POC ₁ represents the difference in POC values between ref_11 and the current picture 610 .

When mv_l1 is derived, the video decoding device may predict the current block 620 based on the first reference block 630 indicated by mv_l0 and the second reference block 640 indicated by mv_l1 (S740).

Depending on the implementation, various implementations proposed by the present invention may use a syntax element indicating enable/disable (eg, linear_MV_coding_enabled_flag) and/or a syntax element indicating a linear relationship of motion (eg, linear_MV_coding_flag or linear_MV_coding_idc) to determine whether to apply to the current Block 620. Here, the syntax element indicating enable/disable may correspond to the above-mentioned enable information, and the syntax element indicating the linear relationship may correspond to the above-mentioned mode information.

linear_MV_coding_enabled_flag is a high-level syntax and may be defined at one or more locations among a sequence level, a picture level, a tile group level, and a slice level. linear_MV_coding_flag may be signaled for each block corresponding to a decoding object.

When linear_MV_coding_enabled_flag=1, whether the embodiment proposed by the present invention is applied can be set for each block by signaling linear_MV_coding_flag for each prediction unit. When linear_MV_coding_flag=1, some or all of motion_info_l1 are not signaled, and motion_info_l1 can be derived using signaled motion_info_l0 (first mode). When linear_MV_coding_flag=0, motion_info_l1 can be signaled as in the conventional method (second mode).

Hereinafter, various embodiments of the present invention will be described on the premise that linear_MV_coding_enabled_flag is defined as activation of the advanced function and linear_MV_coding_flag is set for each block.

Embodiment 3-1

Embodiment 3-1 corresponds to a method in which motion_info_l1 does not signal mvp_l1_flag and mvd_l1 during bidirectional prediction and mvp_l1_flag and mvd_l1 are derived from motion_info_l0 using a linear relationship of motion.

When the second direction is the L0 direction, the motion information in the L0 direction can be derived from the mvd and mvp in the L1 direction and the bidirectional reference picture through the linear relationship of motion. That is, the mvp information and mvd information in direction L0 are not signaled. When the second direction is the L1 direction, the motion information in the L1 direction can be derived from the mvd and mvp in the L0 direction and the bidirectional reference picture through the linear relationship of motion. That is, the mvp information and mvd information in direction L1 are not signaled.

When the motion vector in the direction L1 is derived using a linear relationship (the latter case), the information signaled from the video encoding device to the video decoding device is expressed in syntax as shown in Table 4 below.

[Table 4]

As shown in Table 4, motion_info_10 may be signaled from the video encoding device to the video decoding device by being included in the bitstream. The signaled motion_info_l0 may include ref_idx_l0, mvd_l0, and mvp_l0_flag. ref_idx_l1 may also be signaled by being included in the bitstream. In Embodiment 3-1, the reference pictures (ref_l0 and ref_l1) used to derive mv_l1 correspond to the reference pictures indicated by ref_idx_l0 and ref_idx_l0 signaled from the video encoding device.

When motion_info_l0 is decoded (S910), the video decoding device may infer or derive mv_l0 by using the decoded mvp_l0_flag and mvd_l0 (S920). Equation 1 can be used in this process. Furthermore, ref_idx_l1 can be decoded from the bit stream (S930).

The video decoding device may use linear_MV_coding_enabled_flag to determine whether the motion vector derivation function is activated/deactivated (S940). When linear_MV_coding_enabled_flag indicates activation of the motion vector derivation function, linear_MV_coding_flag may be decoded from the bit stream to determine whether to apply the derivation function proposed by the present invention (S950).

When the decoded linear_MV_coding_flag indicates that the linear relationship of motion is established (S960), the video decoding device can derive mv_l1 on the premise that the linear relationship between mv_l0 and mv_l1 is established (S970). The process of deriving mv_l1 can be achieved by applying each reference picture ref_l0 and ref_l1 and mv_l0 in each direction to Equation 3.

Meanwhile, when the deactivation of the motion vector derivation function is indicated in linear_MV_coding_enabled_flag in operation S940 or the linear_MV_coding_flag does not indicate that a linear relationship of motion is established in operation S960, mv_l1 may be derived through the second mode instead of the first mode. Specifically, the video decoding device may decode mvp_l1_flag and mvd_l1 from the bitstream (S980 and S990), and derive mv_l1 by using mvp_l1_flag and mvd_l1 (S992).

The syntax elements of the above-described embodiment 3-1 are shown in Table 5 below.

[table 5]

9 illustrates that the operation of determining linear_MV_coding_enabled_flag ( S940 ) and the operations of decoding and determining linear_MV_coding_flag ( S950 and S960 ) may be performed after the operation of decoding ref_idx_l1 ( S930 ), but operations S940 to S960 may be performed before the operation of decoding motion_info_l0 ( S910 ).

An example of inferring mv_l1 based on Embodiment 3-1 is illustrated in FIG. 10 . (A) in FIG. 10 and (B) in FIG. 10 each illustrate two types of current picture 610 and reference pictures ref_l0 and ref_l1 according to the size of the POC value in bidirectional prediction. The embodiments to be described below can be applied to both types illustrated in FIG. 10 .

In bidirectional prediction, as shown in (A) of FIG. 10 , the current picture 610 may be located between the reference pictures (ref_l0 and ref_l1) based on POC values (ie, (POC ₀ <POC _cur ) & (POC _cur <POC ₁ )). between. In addition, as shown in (B) of FIG. 10 , based on the POC value (ie, (POC ₀ <POC _cur ) & (POC ₁ <POC _cur )), the bidirectional prediction may include the POC value of the current picture 610 being greater than the reference picture ref_10 and The POC value of ref_l1. Here, POC ₀ indicates the POC value of ref_l0, POC ₁ indicates the POC value of ref_l1, and POC _cur indicates the POC value of the current picture 610.

In both bidirectional predictions, mv_l1 can be derived on the premise of establishing a linear relationship between mv_l0 (solid arrow) and mv_l1 (dashed arrow). In this process, mv_l0 and the reference pictures ref_l0 and ref_l1 in each direction can be used. When mv_l1 is derived, the current block 620 may be predicted based on the reference block 630 indicated by mv_l0 and the reference block 640 indicated by the derived mv_l1.

Embodiment 3-2

Embodiment 3-2 corresponds to a method of inferring mv_l1 based on a linear relationship of motion and then correcting or adjusting mv_l1. Embodiment 3-2 is the same as Embodiment 3-1 in that the motion vector is derived based on the linear relationship of motion, but is different from Embodiment 3-1 in that offset information is used to additionally correct or adjust mv_l1.

The offset information for motion correction corresponds to information indicating the difference between mv_l1 and "adjusted mv_l1". In other words, the offset information corresponds to information indicating the difference between the motion vector derived using the linear relationship of motion (mv_l1) and the measured (actual) motion vector of the current block (adjusted mv_l1).

The offset information may include an offset vector or an offset index. The offset vector corresponds to information indicating the position indicated by "adjusted mv_l1" relative to the position indicated by mv_l1. The offset index corresponds to information obtained by indexing the candidates that may correspond to the offset vector. In the following, each of the two types of offset information will be described by separate implementations.

offset vector

The offset vector can be signaled by also including the offset vector in the bitstream in addition to motion_info_10. As mentioned above, since the offset vector corresponds to the difference between the adjusted mv_l1 and the (unadjusted) mv_l1, the offset vector can be expressed as a motion vector difference (mvd). In addition, since the offset vector corresponds to the difference between the motion vector derived using the motion linear relationship and the measured motion vector of the current block, the offset vector can be compared with the mvd (motion vector of the adjacent block from the motion of the adjacent block) used in the traditional method. The difference between the vector-derived mvp and the mv of the current block). In the present embodiment, information signaled from the video encoding device to the video decoding device for bidirectional prediction is expressed in syntax as shown in Table 6 below.

[Table 6]

In Table 6 above, mvd_l1 can be the mvd or offset vector used in the traditional method. For the current block 620, when a linear relationship of motion is not established, the mvd used in the conventional method may be signaled as mvd_l1, and when a linear relationship of motion is established, the offset vector may be signaled as mvd_l1.

As shown in Table 6, motion_info_l0 may be signaled from the video encoding device to the video decoding device. The signaled motion_info_l0 may include ref_idx_l0, mvd_l0, and mvp_l0_flag as shown in Table 6. Ref_idx_l1 may also be signaled by including ref_idx_l1 in the bitstream.

The video decoding apparatus sets the reference picture indicated by the signaled reference picture information (ref_idx_l0 and ref_idx_l1) as the reference picture (ref_l0 and ref_l1) for inferring mv_l1 (for predicting the current block).

When motion_info_l0 is decoded (S1110), the video decoding device can infer or derive mv_l0 by using mvp_l0_flag and mvd_l0 (S1120). Equation 1 can be used in this process. Also, the video decoding device may decode ref_idx_l1 and mvd_l1 from the bitstream (S1130 and S1140). Here, depending on whether a linear relationship is established, mvd_l1 can correspond to any one of the mvd and offset vector of the traditional method.

The video decoding device may use linear_MV_coding_enabled_flag to determine whether to activate/deactivate the motion vector derivation function (S1150). When linear_MV_coding_enabled_flag indicates activation of the motion vector derivation function, linear_MV_coding_flag may be decoded from the bit stream (S1160).

When linear_MV_coding_flag indicates that a linear relationship of motion is established (S1170), the video decoding apparatus may derive mv_l1 on the premise that a linear relationship of motion is established (S1180). This process may be implemented by applying reference pictures (ref_l0 and ref_l1) and mv_l0 to Formula 3.

The video decoding device may adjust or correct mv_l1 by applying the offset vector (mvd_l1) to the derived mv_l1 (S1182). Specifically, mv_l1 can be adjusted such that the adjusted mv_l1 indicates a position where the offset vector mvd_l1 is translated with the position indicated by mv_l1 as the origin. The adjustment of mv_l1 can be understood as applying the offset vector (mvd_l1) to the assumed predicted motion vector (mvp) under the assumption that the derived mv_l1 is the predicted motion vector (mvp) in the second direction.

Meanwhile, when linear_MV_coding_enabled_flag indicates that the motion vector derivation function is disabled in operation S1150, or linear_MV_coding_flag does not indicate that a linear relationship of motion is established in operation S1170, the video decoding device may derive mv_l1 by a conventional method instead of the derivation method proposed by the present invention. Specifically, the video decoding device may decode mvp_l1_flag (S1190), and derive mv_l1 (S1192) by summing the mvp_l1 indicated by mvp_l1_flag with the mvd_l1 decoded in S1140. Here, mvd_l1 corresponds to the mvd used in the conventional method.

Syntax elements for the above-described embodiments are shown in Table 7 below.

[Table 7]

11 illustrates that the operation of determining linear_MV_coding_enabled_flag ( S1150 ) and the operations of decoding and determining linear_MV_coding_flag ( S1160 and S1170 ) are performed after the operation of decoding mvd_l1 ( S1140 ), but operations S1150 to S1170 may be performed before the operation of decoding motion_info_l0 ( S1110 ).

An example of derivation of mv_l1 based on this embodiment is illustrated in Fig. 12. As shown in Fig. 12, mv_l1 can be derived on the premise that a linear relationship is established between mv_l0 (solid arrow) and mv_l1 (dashed arrow).

Furthermore, assuming that the derived mv_l1 is a predicted motion vector, mv_l1 can be adjusted by moving the position indicated by mv_l1 according to the direction and size indicated by the offset vector mvd_l1. The current block 620 may be predicted based on the reference block 630 indicated by mv_lO and the reference block 640 indicated by the adjusted second motion vector (mv _{A_l1} ).

offset index

The offset index can be signaled by also including the offset index in the bitstream in addition to motion_info_l0. As described above, the offset index corresponds to an index indicating any one of one or more preset offset vector candidates (candidates that may correspond to offset vectors).

In the present embodiment, the information signaled from the video encoding device to the video decoding device for bidirectional prediction is expressed in syntax as shown in Table 8 below.

[Table 8]

In Table 8 above, mv_offset indicates the syntax element corresponding to the offset index. Motion_info_lO may be signaled from the video encoding device to the video decoding device by including motion_info_l0 in the bitstream. The signaled motion_info_l0 may include ref_idx_l0, mvd_l0, and mvp_l0_flag as shown in Table 8. Ref_idx_l1 may also be signaled by including ref_idx_l1 in the bitstream. The video decoding apparatus sets the reference pictures indicated by the signaled reference picture information ref_idx_l0 and ref_idx_l1 as the reference pictures ref_l0 and ref_l1 for inferring mv_l1.

When motion_info_10 is decoded (S1310), the video decoding device can infer or derive mv_10 by using mvp_10_flag and mvd_10 included in motion_info_10 (S1320). Equation 1 can be used in this process. Also, the video decoding device can decode ref_idx_l1 (S1330).

The video decoding device may determine whether to activate or deactivate the motion vector derivation function by analyzing linear_MV_coding_enabled_flag (S1340). When linear_MV_coding_enabled_flag indicates activation of the motion vector derivation function, linear_MV_coding_flag may be decoded from the bit stream (S1350).

When linear_MV_coding_flag indicates that the linear relationship of motion is established (S1360), the video decoding device decodes the offset index mv_offset (S1370), and can derive mv_l1 on the premise that the linear relationship between mv_l0 and mv_l1 is established (S1380). This process can be achieved by applying mv_l0 and bidirectional reference pictures (ref_l0 and ref_l1) to Equation 3.

The video decoding device may adjust or correct mv_l1 by applying the offset vector candidate indicated by the offset index (mv_offset) to the derived mv_l1 (S1382). Specifically, mv_l1 may be adjusted by adding the offset vector candidate indicated by the offset index (mv_offset) to mv_l1. In other words, adjusting mv_l1 can be understood as applying the offset vector candidate indicated by the offset index (mv_offset) to the assumed predicted motion vector under the assumption that the derived mv_l1 is the predicted motion vector (mvp) in the second direction.

Meanwhile, when linear_MV_coding_enabled_flag indicates that the motion vector derivation function is disabled in operation S1340 or linear_MV_coding_flag does not indicate that a linear relationship of motion is established in operation S1360, mv_l1 may be derived by a conventional method rather than the derivation method proposed by the present invention. Specifically, the video decoding apparatus may decode mvd_l1 and mvp_l1_flag from the bitstream (S1390 and S1392), and derive mv_l1 by summing mvp_l1 and mvd_l1 indicated by mvp_l1_flag (S1394).

The syntax elements of the above embodiment are shown in Table 9 below.

[Table 9]

13 illustrates that the operation of determining linear_MV_coding_enabled_flag (S1340) and the operations of decoding and determining linear_MV_coding_flag (S1350 and S1360) are performed after the operation of decoding ref_idx_l1 (S1330), but operations S1340 to S1360 may be performed before the operation of decoding motion_info_l0 (S1310). .

Various types of offset vector candidates used in this embodiment are illustrated in FIG. 14 . (a) in FIG. 14 illustrates offset vector candidates (a circle with an empty interior) when movement of a 4-point offset is allowed. The filled circles represent the mv_l1 derived based on the linear relationship of motion. When movement of 4-point offsets is allowed, a 2-bit fixed length (FL) offset index may be used to indicate any one of the offset vector candidates.

(b) in FIG. 14 illustrates offset vector candidates when movement of 8-point offset is allowed. An 8-point offset vector candidate can be configured by adding four offset vector candidates (circles filled with vertical patterns) to the 4-point offset vector candidate. When movement of an 8-point offset is allowed, a 3-bit fixed-length offset index may be used to indicate any one of the offset vector candidates.

(c) in FIG. 14 illustrates offset vector candidates when movement of 16-point offsets is allowed. The 16-point offset vector candidate can be configured by adding 8 offset vector candidates (circles filled with horizontal patterns) to the 8-point offset vector candidate. When motion of 16-point offsets is allowed, a 4-bit fixed-length offset index may be used to indicate any one of the offset vector candidates.

(d) in FIG. 14 illustrates another example of a case where movement of 16-point offset is allowed. The 16-point offset vector candidate can be configured by combining an 8-point offset vector candidate filled with a horizontal pattern and an 8-point offset vector candidate filled with a diagonal pattern. When motion of 16-point offsets is allowed, a 4-bit fixed-length offset index may be used to indicate any one of the offset vector candidates.

The various types of offset vector candidates described with reference to FIG. 14 may be determined or defined at one or more locations of the picture level header, tile group header, tile header, and/or CTU header. Which type of. That is, the shape of the offset vector candidate can be determined using information (identification information) signaled from the video encoding device, and the identification information can be defined at the various positions described above. Since the identification information determines or identifies any one of various types of offset vector candidates, the number of offset vector candidates, the size of each candidate, and the direction of each candidate can be determined by the identification information.

In addition, which type of various types of offset vector candidates is set can be determined in advance by using the same rule at the video encoding device and the video decoding device.

Fourth embodiment

The fourth embodiment corresponds to a method of deriving the motion in a direction establishing a linear relationship among the horizontal and vertical motion directions without signaling using motion_info_10 while adjusting using additional signaled information (offset information) Movement in directions for which a linear relationship is not established.

For example, when a linear relationship is established only for the horizontal axis component of motion, the derived horizontal axis component of mv_l1 can be used without modification, but the vertical axis for which no linear relationship is established is adjusted by using additional signaled offset information. axis component. As another example, when a linear relationship is established only for the vertical axis component of motion, the derived vertical axis component of mv_l1 is used without modification, but no linear relationship is established by adjusting using the additional signaled offset information the horizontal axis component of .

The fourth embodiment can be implemented in combination with the above-mentioned embodiment 3-1 or 3-2. Hereinafter, a combination of the fourth embodiment and Embodiment 3-1 and a combination of the fourth embodiment and Embodiment 3-2 will be described respectively.

Embodiment 4-1

Embodiment 4-1 corresponds to a form in which the fourth embodiment is combined with Embodiment 3-1. In this embodiment, information signaled from the video encoding device to the video decoding device for bidirectional prediction is expressed in syntax as shown in Table 10 below.

[Table 10]

In Table 10, mvd_l1 can be offset information (offset vector) or mvd of the traditional method. For example, when no linear relationship is established for the horizontal axis component, mvd_l1 is the offset vector of the horizontal axis component, and when no linear relationship is established for the vertical axis component, mvd_l1 may be the offset vector of the vertical axis component. Furthermore, when a linear relationship is not established for both the horizontal axis component and the vertical axis component, mvd_l1 may be mvd of the conventional method. When a linear relationship is established for both the horizontal axis component and the vertical axis component, mvd_l1 is not signaled.

motion_info_10 may be signaled from the video encoding device to the video decoding device by being included in the bitstream. The signaled motion_info_l0 may include ref_idx_l0, mvd_l0, and mvp_l0_flag. Ref_idx_l1 may also be signaled by including ref_idx_l1 in the bitstream. The video decoding apparatus sets the reference picture indicated by the signaled reference picture information (ref_idx_l0 and ref_idx_l1) as the reference picture (ref_l0 and ref_l1) for inferring mv_l1.

When motion_info_l0 is decoded (S1510), the video decoding device can infer or derive mv_l0 by using mvp_l0_flag and mvd_l0 (S1520). Equation 1 can be used in this process. Also, the video decoding device may decode ref_idx_l1 from the bit stream (S1530).

When linear_MV_coding_enabled_flag indicates activation of the motion vector derivation function (S1540), the video decoding device decodes linear_MV_coding_idc from the bit stream (S1550). Here, linear_MV_coding_idc is information indicating whether motion has a linear relationship, and the component in which a linear relationship is established among the horizontal axis component and the vertical axis component of motion can be indicated by using this information.

When linear_MV_coding_idc=none (S1560), since no linear relationship is established for both components, mvp_l1_flag and mvd_l1 are signaled as in the conventional method. Therefore, the video decoding device can decode mvp_l1_flag and mvd_l0 from the bit stream (S1562), and derive mv_l1 by using the decoded information (S1564). Also, when linear_MV_coding_enabled_flag does not indicate activation of the motion vector derivation function in operation S1540, the video decoding device may derive mv_l1 by using the decoded mvp_l1_flag and mvd_l1 (S1562 and S1564).

When linear_MV_coding_idc=x (S1570), since a linear relationship is established only for the horizontal axis component (x), the offset vector (mvd_l1, y) for the vertical axis component (y) for which a linear relationship is not established is signaled. Therefore, the video decoding device decodes the offset vector (mvd_l1, y) for the vertical axis component (S1572), and derives mv_l1 using a linear relationship. Furthermore, the video decoding device may adjust mv_l1 by applying the offset vector (mvd_l1, y) for the vertical axis component to the derived mv_l1 (S1576).

The video decoding device may use the "derived mv_l1" for the horizontal axis component without modification, and use the adjusted second motion vector (mvA_l1) for the vertical axis component. The derived horizontal axis component of mv_l1 and the horizontal axis component of the adjusted second motion vector (mvA_l1) may be the same.

When linear_MV_coding_idc=y (S1580), since the linear relationship is established only for the vertical axis component, the offset vector (mvd_l1,x) for the horizontal axis component for which the linear relationship is not established is signaled. Therefore, the video decoding device can decode the offset vector (mvd_l1,x) for the horizontal axis component (S1582), and apply the offset vector (mvd_l1,x) for the horizontal axis component to the derived mv_l1(S1584) to adjust mv_l1(S1586).

The video decoding apparatus may use the "derived mv_l1" for the vertical axis component without modification, and use the adjusted second motion vector (mvA_l1) for the horizontal axis component. The derived vertical axis component of mv_l1 and the vertical axis component of the adjusted second motion vector (mvA_l1) may be the same.

When linear_MV_coding_idc=(x&y) (S1580), since a linear relationship is established for both the horizontal axis component and the vertical axis component, mvd_l1 (offset information in the second direction or mvd information) is not signaled. In this case, the video decoding device derives mv_l1 by using motion_info_l0 and ref_idx_l1 (S1590).

The syntax elements of Embodiment 4-1 are shown in Table 11 below.

[Table 11]

15 illustrates that the operation of determining linear_MV_coding_enabled_flag ( S1540 ) and the operations of decoding and determining linear_MV_coding_idc ( S1550 to S1580 ) may be performed after the operation of decoding ref_idx_l1 ( S1530 ), but operations S1540 to S1580 may be performed before the operation of decoding motion_info_l0 ( S1510 ).

Implementation 4-2

Embodiment 4-2 corresponds to a combination of the fourth embodiment and Embodiment 3-2. In this embodiment, the information signaled from the video encoding device to the video decoding device for bidirectional prediction is expressed in syntax as shown in Table 10 above.

In Table 10, mvd_l1 can be offset information (offset vector) or mvd of the traditional method. For example, mvd_l1 is an offset vector for the horizontal axis component when no linear relationship is established for the horizontal axis component, and mvd_l1 may be an offset vector for the vertical axis component when no linear relationship is established for the vertical axis component. Furthermore, when a linear relationship is not established for both the horizontal axis component and the vertical axis component, mvd_l1 may be mvd of the conventional method. When a linear relationship is established for both the horizontal axis component and the vertical axis component, mvd_l1 may be the offset vector for both components.

motion_info_10 may be signaled from the video encoding device to the video decoding device by being included in the bitstream. The signaled motion_info_l0 may include ref_idx_l0, mvd_l0, and mvp_l0_flag. Ref_idx_l1 may also be signaled by including ref_idx_l1 in the bitstream. The video decoding apparatus sets the reference picture indicated by the signaled reference picture information (ref_idx_l0 and ref_idx_l1) as the reference picture (ref_l0 and ref_l1) for inferring mv_l1.

When motion_info_l0 is decoded (S1610), the video decoding device can infer or derive mv_l0 by using mvp_l0_flag and mvd_l0 (S1620). Equation 1 can be used in this process. In addition, the video decoding device may decode ref_idx_l1 from the bit stream (S1630).

When linear_MV_coding_enabled_flag indicates activation of the motion vector derivation function (S1640), the video decoding device decodes linear_MV_coding_idc from the bit stream (S1650).

When linear_MV_coding_idc=none (S1660), since no linear relationship is established for the two components, mvp_l1_flag and mvd_l1 are signaled as in the conventional method. Therefore, the video decoding device can decode mvp_l1_flag and mvd_l1 from the bitstream (S1662) and derive mv_l1 by using the decoded information (S1664). Even when linear_MV_coding_enabled_flag does not indicate activation of the motion vector derivation function in operation S1640, the video decoding device may derive mv_l1 by using the decoded mvp_l1_flag and mvd_l1 (S1662 and S1664).

When linear_MV_coding_idc=x (S1670), since the linear relationship is established only for the horizontal axis component, the offset vector (mvd_l1, y) for the vertical axis component for which the linear relationship is not established is signaled. Therefore, the video decoding device decodes the offset vector (mvd_l1, y) for the vertical axis component (S1672), and derives mv_l1 using a linear relationship (1674). Then, the video decoding device may adjust mv_l1 by applying the offset vector (mvd_l1,y) for the vertical axis component to the derived mv_l1 (S1676).

The video decoding device may use the "derived mv_l1" without change for the horizontal axis component, and use the adjusted second motion vector (mvA_l1) for the vertical axis component. The horizontal axis component of the derived mv_l1 and the horizontal axis component of the adjusted second motion vector (mvA_l1) may be the same.

When linear_MV_coding_idc=y (S1680), since the linear relationship is established only for the vertical axis component, the offset vector (mvd_l1,x) for the horizontal axis component for which the linear relationship is not established is signaled. Therefore, the video decoding device can decode the offset vector (mvd_l1, x) for the horizontal axis component (S1682), derive mv_l1 derived by using the linear relationship (S1684), and convert the offset vector (mvd_l1 for the horizontal axis component , x) is applied to the derived mv_l1 to adjust mv_l1 (S1686).

The video decoding device may use the "derived mv_l1" for the vertical axis component without modification and use the adjusted second motion vector (mvA_l1) for the horizontal axis component. The derived vertical axis component of mv_l1 and the vertical axis component of the adjusted second motion vector (mvA_l1) may be the same.

When linear_MV_coding_idc=(x&y) (S1680), since a linear relationship is established for both the horizontal axis component and the vertical axis component, the offset vector (mvd_l1, for both the horizontal axis component and the vertical axis component) is signaled x and y). Therefore, the video decoding device decodes the offset vector (mvd_l1, x, and y) for both the horizontal axis component and the vertical axis component from the bit stream (S1690), and can apply the offset vector (mvd_l1, x, and y) by For mv_l1(S1692)(S1694) derived using a linear relationship.

The syntax elements of Embodiment 4-2 are shown in Table 12 below.

[Table 12]

16 illustrates that the operation of determining linear_MV_coding_enabled_flag (S1640) and the operation of decoding and determining linear_MV_coding_idc (S1650 to S1680) may be performed after the operation of decoding ref_idx_l1 (S1630), but the operations S1640 to S1640 may be performed before the operation of decoding motion_info_l0 (S1610). S1680.

An example of deriving mv_l1 based on the fourth embodiment is illustrated in FIG. 17 . The example shown in FIG. 17 corresponds to an example in which a linear relationship is established with respect to the vertical axis component.

As shown in Figure 17, mv_l1 can be derived on the premise that a linear relationship is established between mv_l0 (solid arrow) and mv_l1 (dashed arrow).

Since no linear relationship is established for the horizontal axis component, mv_l1 can be adjusted by moving the position indicated by the derived mv_l1 in the horizontal axis direction according to the magnitude indicated by the offset vector mvd_l1. The final motion vector mv _{A_l1} in the second direction can be derived by using the vertical axis component of the unmodified mv_l1 and the adjusted horizontal axis component of the second motion vector mvA_l1. The current block 620 may be predicted based on the reference block 630 indicated by mv_lO and the reference block 640 indicated by the adjusted second motion vector (mv _{A_l1} ).

Fifth embodiment

The fifth embodiment corresponds to a method of using a preset reference picture as a reference picture for deriving mv_l1. The preset reference picture refers to the preset reference picture to be used when establishing a motion linear relationship.

In the fifth embodiment, the reference picture information (ref_idx_l0 and ref_idx_l1) is not signaled in units of blocks, but may be signaled at a high level. Here, the high level may correspond to one or more of a picture level header, a tile group level header, a slice header, a tile header, and/or a CTU header. The preset reference picture may be referred to as a "representative reference picture" or a "linear reference picture", and the reference picture information signaled at a high level may be referred to as "representative reference picture information" or "linear reference picture information". When a linear relationship of motion is established, a preset linear reference picture is used in units of blocks.

The linear reference picture information signaled in the tile group header is shown in Table 13 below.

[Table 13]

In Table 13, each of linear_ref_idx_l0 and linear_ref_idx_l1 represents linear reference picture information signaled for each direction.

FIG. 18 illustrates an example of a method of specifying a reference picture by signaling reference picture information for each block in a conventional method or a method of specifying a linear reference picture by the method proposed in the present invention.

The linear reference picture information (linear_ref_idx_l0 and linear_ref_idx_l1) can be signaled to the video decoding device from the video encoding device at a high level. The video decoding device may set the linear reference picture (linear_ref_l0 and linear_ref_l1) by selecting the reference picture indicated by the signaled linear reference picture information (linear_ref_idx_l0 and linear_ref_idx_l1) within the reference picture list.

When linear_MV_coding_enabled_flag indicates activation of the motion vector derivation function (S1810), the video decoding device decodes linear_MV_coding_flag from the bit stream (S1820).

When linear_MV_coding_flag indicates that a linear relationship of motion is established (S1830), the video decoding device may use preset linear reference pictures (linear_ref_l0 and linear_ref_l1) to derive reference pictures (ref_l0 and ref_l1) for deriving mv_l1 (S1840 and S1850). That is, the preset linear reference pictures (linear_ref_l0 and linear_ref_l1) can be set as the reference pictures (ref_l0 and ref_l1).

Meanwhile, when linear_MV_coding_enabled_flag does not indicate activation of the motion vector derivation function in operation S1810 or linear_MV_coding_flag does not indicate establishment of a linear relationship of motion in operation S1830, reference picture information (ref_idx_l0 and ref_idx_l1) may be signaled. The video decoding apparatus may decode the reference picture information (ref_idx_l0 and ref_idx_l1) (S1860 and S1870) and set a reference picture using the reference picture information.

The method for setting a reference picture proposed by the present invention can be implemented in combination with the above-mentioned embodiments. FIG. 19 illustrates a form that combines the method of setting a reference picture proposed by the present invention and the above-mentioned Embodiment 3-1.

For the first direction, when linear_MV_coding_enabled_flag indicates activation of the motion vector derivation function (S1910), decode linear_MV_coding_flag (S1920). When linear_MV_coding_flag indicates that a linear relationship of motion is established, a preset linear reference picture (linear_ref_l0) can be derived as a reference picture (ref_l0) (S1940). On the other hand, when linear_MV_coding_enabled_flag does not indicate activation of the motion vector derivation function or linear_MV_coding_flag does not indicate establishment of a linear relationship of motion, a reference picture (ref_l0) can be set by using reference picture information (ref_idx_l0) decoded from a bitstream (S1962).

When the derivation or setting of the reference picture for the first direction is completed, mvd_l0 and mvp_l0_flag are decoded (S1950), and mv_l0 can be derived using the decoded information (S1960).

For the second direction, when linear_MV_coding_flag indicates that a linear relationship of motion is established (S1970), a preset linear reference picture (linear_ref_l1) may be used to derive or set the reference picture (ref_l1) (S1972). On the other hand, when linear_MV_coding_flag does not indicate establishing a linear relationship of motion, the reference picture (ref_l1) may be set using the reference picture information (ref_idx_l1) decoded from the bitstream (S1974).

When the derivation or setting of the reference picture for the second direction is completed, when the linear relationship of linear_MV_coding_flag indicating motion is established (S1980), mv_l1 having a linear relationship with mv_l0 can be derived (S1982). On the other hand, when linear_MV_coding_flag does not indicate that the linear relationship of motion is established (S1980), mv_l1 can be derived using mvd_l1 and mvp_l1_flag (S1990 and S1992) decoded from the bit stream (S1994).

The syntax elements of the above embodiment are shown in Table 14 below.

[Table 14]

Figure 20 illustrates a combination of the method for setting a reference picture proposed by the present invention and the above-mentioned Embodiment 3-2.

For the first direction, when linear_MV_coding_enabled_flag indicates activation of the motion vector derivation function (S2010), linear_MV_coding_flag is decoded (S2020). When the linear_MV_coding_flag indicates that a linear relationship of motion is established (S2030), the reference picture (ref_l0) can be derived or set using the preset linear reference picture (linear_ref_l0) (S2040). On the other hand, when linear_MV_coding_enabled_flag does not indicate activation of the motion vector derivation function (S2010) or linear_MV_coding_flag does not indicate that a linear relationship of motion is established (S2030), the reference picture information (ref_idx_l0) decoded from the bit stream may be used to set the reference picture ( ref_l0)(S2062).

When the derivation or setting of the reference picture for the first direction is completed, mvd_l0 and mvp_l0_flag are decoded (S2050), and mv_l0 can be derived using the decoded information (S2060).

For the second direction, when linear_MV_coding_flag indicates that a linear relationship of motion is established (S2070), a preset linear reference picture (linear_ref_l1) may be used to derive or set the reference picture (ref_l1) (S2072). On the other hand, when linear_MV_coding_flag does not indicate that the linear relationship of motion is established, the reference picture (ref_l1) may be set using the reference picture information (ref_idx_l1) decoded from the bitstream (S2074).

When the derivation or setting of the reference picture for the second direction is completed, mvd_l1 is decoded from the bit stream (S2080), and mvd_l1 corresponds to the offset vector or the mvd of the conventional method, as in Embodiment 3-2.

When linear_MV_coding_flag indicates that the linear relationship of motion is established (S2090), mv_l1 having a linear relationship with mv_l0 is derived (S2092), and mv_l1 can be adjusted by applying the offset vector (mvd_l1) to the derived mv_l1 (S2094). On the other hand, when linear_MV_coding_flag does not indicate that the linear relationship of motion is established (S2090), mv_l1 may be derived using mvp_l1_flag decoded from the bit stream (S2096 and S2098). In this process, mvp_l1 indicated by mvp_l1_flag and decoded mvd_l1 (mvd of the traditional method) can be used.

Although exemplary embodiments of the present invention have been described for illustrative purposes, those skilled in the art will appreciate that various modifications and variations are possible without departing from the spirit and scope of the invention. Example implementations have been described for the sake of brevity and clarity. Accordingly, one of ordinary skill in the art will understand that the scope of the present invention is not limited by the specifically described embodiments above, but includes the appended claims and their equivalents.

Cross-references to related applications

This application claims priority to Patent Application No. 10-2018-0171254 filed in Korea on December 27, 2018 and Patent Application No. 10-2019-0105769 filed in Korea on August 28, 2019, the entire contents of which are incorporated herein by reference.

Claims (11)

1. A method of inter-predicting the current block using any one of multiple bidirectional prediction modes, the method includes the following steps: decoding enablement information indicating whether a first mode of the plurality of bidirectional prediction modes is allowed; When the enabling information indicates that the first mode is allowed, decoding mode information at the block level of the current block in the bitstream, the mode information indicating whether to apply the first mode to the current block; When the mode information indicates that the first mode is applied to the current block, decoding, from the bitstream, first motion information including differential motion vector information and predicted motion vector information for the first motion vector and second motion information excluding at least a portion of the differential motion vector information and predicted motion vector information for the second motion vector; and deriving the first motion vector based on the first motion information, and deriving the second motion vector based on at least a portion of the first motion information and based on the second motion information; and predicting the current block using a reference block in a first reference picture indicated by the first motion vector and a reference block in a second reference picture indicated by the second motion vector, Among them, in the first mode, the first reference picture and the second reference picture are determined at a higher level than the block level. 2. The method of claim 1, further comprising the step of: when the mode information indicates that the first mode is not to be applied to the current block, Decoding the first motion information and third motion information including the differential motion vector information and the predicted motion vector information for the second motion vector from the bitstream; and The first motion vector is derived based on the first motion information and the second motion vector is derived based on the third motion information. 3. The method of claim 1, wherein when the enabling information indicates that the first mode is not activated, the mode information is not decoded from the bitstream and the mode information is set to indicate that no application The first mode. 4 . The method of claim 1 , wherein the enabling information is decoded at a sequence level, a picture level, a tile group level, or a slice level. 5. The method of claim 1, wherein the high level is a picture level, a tile group level, a slice level, a tile level or a coding tree unit level. 6. The method of claim 1, wherein the first reference picture and the second reference picture are determined based on a Picture Order Count POC difference between a reference picture included in a reference picture list and a current picture. 7. The method of claim 1, further comprising the step of applying offset information included in the bitstream to the second motion vector after deriving the second motion vector. adjust the second motion vector, wherein the current block is predicted by using a reference block indicated by the adjusted second motion vector in the second reference picture and a reference block indicated by the first motion vector in the first reference picture . 8. The method of claim 7, wherein the offset information is an offset vector with a position indicated by the second motion vector as an origin, and The adjusting includes adjusting the second motion vector to a position indicated by the offset vector. 9. The method of claim 7, wherein the offset information is an offset index indicating any one of a plurality of preset offset vector candidates, and The adjusting includes adjusting the second motion vector by applying an offset vector candidate indicated by the offset index to the second motion vector. 10. A video encoding method that uses any one of a plurality of bidirectional prediction modes to perform inter prediction on the current block, the method comprising the following steps: encoding enabling information indicating whether a first mode of the plurality of bidirectional prediction modes is allowed; When the enabling information indicates that the first mode is allowed, encoding mode information at the block level of the current block, the mode information indicating whether to apply the first mode to the current block; and When the mode information indicates that the first mode is applied to the current block, encoding and signaling first motion information including differential motion vector information and predicted motion vector information for the first motion vector and second motion information excluding at least a portion of the differential motion vector information and predicted motion vector information for the second motion vector, and encoding a residual block, the residual block being a difference between the current block and a prediction block of the current block, wherein the prediction block is generated by using a reference block indicated by the first motion vector in a first reference picture and a reference block indicated by the second motion vector in a second reference picture, Wherein, in the first mode, the first reference picture and the second reference picture are determined at a higher level than the block level. 11. A method for transmitting a bitstream containing encoded video data, the method comprising the steps of: generating the bitstream by encoding the current block using any one of a plurality of bidirectional prediction modes; and transmit said bitstream to a video decoding device, Wherein, generating the bitstream includes: encoding enabling information indicating whether a first mode of the plurality of bidirectional prediction modes is allowed; When the enabling information indicates that the first mode is allowed, encoding mode information at the block level of the current block, the mode information indicating whether to apply the first mode to the current block; and When the mode information indicates that the first mode is applied to the current block, encoding first motion information including differential motion vector information and predicted motion vector information for the first motion vector and second motion information excluding at least a portion of differential motion vector information and predicted motion vector information for the second motion vector, and encoding a residual block, the residual block being a difference between the current block and a prediction block of the current block, wherein the prediction block is generated by using a reference block indicated by the first motion vector in a first reference picture and a reference block indicated by the second motion vector in a second reference picture, Wherein, in the first mode, the first reference picture and the second reference picture are determined at a higher level than the block level.