Method and apparatus for video coding using nonrectangular block splitting structure
A method and an apparatus are disclosed for video coding using a nonrectangular block splitting structure prediction and transform accordingly. The video coding method and the apparatus additionally apply a nonrectangular block splitting structure, in addition to an existing splitting structure, to a current block. The video coding method and the apparatus generate reference samples according to the applied splitting structure and effectively perform prediction and transform on split blocks.
1 . A method for decoding a block having a size of W×H, performed by a video decoding apparatus, the method comprising:
obtaining a nonrectangular splitting flag from a bitstream for the block having the size of W×H, wherein W is a width of the block, H is a height of the block, and the nonrectangular splitting flag indicates nonrectangular block splitting for the block; and
checking the nonrectangular splitting flag,
wherein, in response to a determination that the nonrectangular splitting flag is true, the method further includes
obtaining a nonrectangular splitting index from the bitstream, wherein the nonrectangular splitting index indicates one of nonrectangular splitting methods,
determining prediction units by splitting the block having the size of W×H using a nonrectangular splitting method indicated by the nonrectangular splitting index,
obtaining intra prediction modes for the prediction units from the bitstream, and
generating a prediction block of the block using the intra prediction modes.
2 . The method of claim 1 , wherein generating the prediction block includes:
generating reference samples of each prediction unit from previously reconstructed blocks based on each intra prediction mode;
generating prediction samples of each prediction unit using the reference samples; and
correcting the prediction samples of each prediction unit.
3 . The method of claim 1 , wherein the nonrectangular splitting method includes:
generating blocks B 1 , B 2 , . . . , B K (here, K is a natural number of 2 or greater) as the prediction units using one of nonrectangular block splitting structure among SPLIT_RD2LT, SPLIT_LD2RT, SPLIT_RT2LD, and SPLIT_LT2RD,
wherein each of the SPLIT_RD2LT, SPLIT_LD2RT, SPLIT_RT2LD, and SPLIT_LT2RD includes block B 1 located at a bottom right, bottom left, top right, and top left of the block having the size of W×H.
4 . The method of claim 3 , wherein, for the block having the size of W×H in which 1 or more natural numbers N 1 , N 2 , . . . , N K satisfy N 1 +N 2 + . . . +N K =W and 1 or more natural numbers M 1 , M 2 , . . . , M K satisfy M 1 +M 2 + . . . +M K =H, the prediction units are generated by excluding blocks B 1 , B 2 , . . . , B k−1 from a rectangular block having a size of (N 1 +N 2 + . . . +N k )×(M 1 +M 2 + . . . +M k ) in which a rectangular block B 1 having a size of N 1 ×M 1 is located on the bottom right, the bottom left, the top right, or the top left of the block having the size of W×H and a nonrectangular block Bk corresponding to k (where 2≤k≤K) is located on the bottom right, the bottom left, the top right, or the top left of the block having the size of W×H.
5 . The method of claim 3 , wherein the method further includes:
deriving a prediction order in an ascending order or a descending order according to the nonrectangular splitting method,
wherein the ascending order indicates a prediction order of B 1 →B 2 → . . . →B K and the descending order indicates a prediction order of B K →B K−1 → . . . →B 1 .
6 . The method of claim 1 , wherein obtaining the intra prediction modes includes:
limiting an available intra prediction mode according to the nonrectangular splitting method, a size of the prediction units, and an aspect ratio of the prediction units.
7 . The method of claim 2 , wherein generating the reference samples includes:
generating the reference samples using reconstructed samples existing in neighboring positions of the prediction units and neighboring samples of the block having the size of W×H,
wherein when a reference sample does not exist in a neighboring position of the prediction units, the reference sample is copied from a sample value in a position closest to the neighboring position.
8 . The method of claim 7 , wherein generating the reference samples includes:
compensating the reference samples of the prediction units by calculating difference signals between adjacent samples using the neighboring samples of the block having the size of W×H and adding the difference signals to copied samples among the reference samples of the prediction units.
9 . The method of claim 2 , wherein generating the reference samples includes:
when a reference sample does not exist in a neighboring position of the prediction units, generating a reference sample in the neighboring position by weighting the sample values in positions closest to the neighboring position.
10 . The method of claim 2 , wherein correcting the prediction samples of each prediction unit includes:
using samples existing in reconstructed available rows and columns closest to the each prediction unit to correct the prediction samples of each prediction unit.
11 . The method of claim 2 , wherein correcting the prediction samples of each prediction unit includes:
for block BN (1≤N≤K) among the blocks B 1 , B 2 , . . . , B K , using 3+2(K−N) samples to correct a prediction sample included in the block BN.
12 . The method of claim 1 , in response to the determination that the nonrectangular splitting flag is true, the method further comprising:
generating residual signals of the prediction units; and
generating a reconstructed block by adding the prediction block and the residual signals.
13 . The method of claim 12 , wherein generating the residual signals includes:
obtaining inverse quantized signals of the prediction units and a transform unit splitting method from the bitstream, wherein the transform unit splitting method indicates one of a plurality of transform unit configurations when the prediction units are nonrectangular;
generating a transform unit according to the transform unit splitting method; and
generating residual signals by inversely transforming the inverse quantized signals using the transform unit when the prediction units are nonrectangular.
14 . The method of claim 13 , wherein generating the transform unit includes:
applying a same transform unit configuration to nonrectangular prediction units among the prediction units.
15 . A method for encoding a block having a size of W×H, performed by a video encoding apparatus, the method comprising:
determining whether to perform nonrectangular block splitting for the block having the size of W×H, wherein W is a width of the block, and His a height of the block; and
wherein, in response to a determination that the nonrectangular block splitting is performed, the method further includes
determining one of nonrectangular splitting methods,
determining prediction units by splitting the block having the size of W×H using the determined nonrectangular splitting method,
determining intra prediction modes for the prediction units,
generating a prediction block of the block using the intra prediction modes, and
encoding a nonrectangular splitting flag indicating nonrectangular block splitting for the block and a nonrectangular splitting index indicating the determined nonrectangular splitting method.
16 . The method of claim 15 , wherein generating the prediction block includes:
generating reference samples of each prediction units from previously reconstructed blocks based on each intra prediction mode;
generating prediction samples of each prediction unit using the reference samples; and
correcting the prediction samples of each prediction unit.
17 . The method of claim 15 , in response to the determination that the nonrectangular block splitting is performed, the method further comprising:
generating residual signals by subtracting the prediction block from an original block of the block; and
transforming the residual signals.
18 . The method of claim 17 , wherein transforming the residual signals includes:
obtaining a transform unit splitting method, wherein the transform unit splitting method indicates one of a plurality of transform unit configurations when the prediction units are nonrectangular;
generating a transform unit according to the transform unit splitting method; and
transforming the residual signals using the transform unit when the prediction units are nonrectangular.
19 . The method of claim 18 , wherein generating the transform unit includes:
applying a same transform unit configuration to nonrectangular prediction units among the prediction units.
20 . A method for transmitting a bitstream containing encoded video data, the method comprising:
generating the bitstream for a current block in an image; and
transmitting the bitstream,
wherein generating the bitstream for the image comprises:
determining whether to perform nonrectangular block splitting for a block having a size of W×H, wherein W is a width of the block, and H is a height of the block, and
wherein, in response to a determination that the nonrectangular block splitting is performed, the one or more processors are further configured to
determine one of nonrectangular splitting methods,
determine prediction units by splitting the block having the size of W×H using the determined nonrectangular splitting method,
determine intra prediction modes for the prediction units,
generate a prediction block of the block using the intra prediction modes, and
encode a nonrectangular splitting flag indicating nonrectangular block splitting for the block and a nonrectangular splitting index indicating the determined nonrectangular splitting method.