Back to results
US 2023/0156239 A1Published application
Apparatus and method for video encoding or decoding
AssigneeSK Telecom Co., Ltd.
InventorsJeong-yeon LIM, Jae-seob SHIN, Se-hoon SON, Sun-young LEE
Life of this patent
Estimating the term…
Abstract
Disclosed herein are a QTBT split structure allowing blocks of various shapes capable of more efficiently reflecting various local characteristics of video and a method of efficiently signaling the split structure.
Claims7
A method for decoding video data, comprising:
- receiving a bitstream including a coding tree unit (CTU) of encoded video data and split information related to a tree split structure of the CTU, wherein the bitstream includes information on maximum block size allowed for a root node of a binary tree;
- determining a square or rectangular shaped coding block to be decoded, wherein the coding block is a block which is split from the CTU in the tree split structure using the split information and corresponds to a leaf node of the tree split structure, the tree split structure allowing, depending on the split information, a quadtree having multi-depth of consecutive quadtree divisions and a binary tree having multi-depth of consecutive binary tree divisions, wherein the tree split structure is configured such that the binary tree is rooted from a leaf node of the quadtree when a size of a block corresponding to the leaf node of the quadtree is not greater than the maximum block size;
- after determining the coding block to be decoded by splitting the CTU in the tree split structure, determining a prediction partition mode for the coding block corresponding to the leaf node of the tree split structure among a plurality of prediction partition modes using mode type information included in the bitstream, the plurality of prediction partition modes including at least one prediction partition mode allowing the coding block to be predicted by being split into two equal-sized triangles;
- and
- decoding the coding block corresponding to the leaf node according to the determined prediction partition mode,
- wherein, when the coding block is split into the two triangles according to the determined prediction partition mode, the decoding of the coding block comprises:
- configuring a plurality of merge candidates comprised of motion vectors of pre-decoded blocks neighboring to the coding block;
- reconstructing, from the bitstream, two merge indexes which respectively correspond to the two triangles, wherein each of the merge indexes is used for indicating one of the plurality of merge candidates;
- setting motion vectors respectively corresponding to the two triangles equal to merge candidates which are respectively indicated by the two merge indexes among the plurality of merge candidates;
- and
- inter-predicting the coding block which is split into the two triangles, by using the set motion vectors, and
- wherein the at least one prediction partition mode includes a down-right diagonal split type and an up-right diagonal split type.
9 more elements
An apparatus of encoding video data, the apparatus comprising at least one processor configured to:
- encode restriction information related to a tree split structure, the restriction information including information on maximum block size allowed for a root node of a binary tree;
- split a coding tree unit (CTU) in the tree split structure and encode split information related to the tree split structure of the CTU, the tree split structure allowing a quadtree having multi-depth of consecutive quadtree divisions and a binary tree having multi-depth of consecutive binary tree divisions, wherein the tree split structure is configured such that the binary tree is rooted from a leaf node of the quadtree when a size of a block corresponding to the leaf node of the quadtree is not greater than the maximum block size;
- select, from a plurality of prediction partition modes, a prediction partition mode for a square or rectangular shaped coding block corresponding to a leaf node of the tree split structure, and encode mode type information for representing the selected prediction partition mode, wherein the plurality of prediction partition modes includes at least one prediction partition mode allowing the coding block to be predicted by being split into two equal-sized triangles;
- and
- encode the coding block according to the selected prediction partition mode,
- wherein, when the coding block is split into the two triangles, the processor is configured to encode the coding block by performing a process comprising:
- configuring a plurality of merge candidates comprised of motion vectors of pre-decoded blocks neighboring to the coding block;
- setting motion vectors respectively corresponding to the two triangles equal to two merge candidates included in the plurality of merge candidates;
- encoding merge index information for indicating the two merge candidates which are set as motion vectors respectively corresponding to the two triangles,
- inter-predicting the coding block which is split into the two triangles, by using the set motion vectors,
- wherein the at least one prediction partition mode includes a down-right diagonal split type and an up-right diagonal split type.
8 more elements
A non-transitory computer readable medium storing a bitstream generated by a method of encoding video data, the method comprising:
- encoding restriction information related to a tree split structure, the restriction information including information on maximum block size allowed for a root node of a binary tree;
- splitting a coding tree unit (CTU) in a tree split structure and encoding split information related to the tree split structure of the CTU, the tree split structure allowing a quadtree having multi-depth of consecutive quadtree divisions and a binary tree having multi-depth of consecutive binary tree divisions, wherein the tree split structure is configured such that the binary tree is rooted from a leaf node of the quadtree when a size of a block corresponding to the leaf node of the quadtree is not greater than the maximum block size;
- selecting, from a plurality of prediction partition modes, a prediction partition mode for a square or rectangular shaped coding block corresponding to a leaf node of the tree split structure, and encoding mode type information for representing the selected prediction partition mode, wherein the plurality of prediction partition modes includes at least one prediction partition mode allowing the coding block to be predicted by being split into two equal-sized triangles;
- and
- encoding the coding block according to the selected prediction partition mode,
- wherein, when the coding block is split into the two triangles, the encoding of the coding block comprises:
- configuring a plurality of merge candidates comprised of motion vectors of pre-decoded blocks neighboring to the coding block;
- setting motion vectors respectively corresponding to the two triangles equal to two merge candidates included in the plurality of merge candidates;
- encoding merge index information for indicating the two merge candidates which are set as motion vectors respectively corresponding to the two triangles,
- inter-predicting the coding block which is split into the two triangles, by using the set motion vectors,
- wherein the at least one prediction partition mode includes a down-right diagonal split type and an up-right diagonal split type.
8 more elements
Dependencies are inferred from the claim text.