Inter-prediction method and apparatus in image coding system
An image decoding method performed by a decoding apparatus according to the present invention is characterised by comprising the steps of: deriving a reference picture list 0 (L 0 ) and a reference picture list 1 (L 1 ); deriving two motion vectors (MV) for a current block, wherein the two MVs include MVL 0 for the L 0 and MVL 1 for the L 1 ; determining whether to apply BIO prediction that derives motion vectors for each sample in the current block based on the MVL 0 and the MVL 1 ; deriving motion vectors for each sample in the current block based on the MVL 0 and the MVL 1 if it is determined to apply the BIO prediction; and deriving a prediction sample based on the motion vectors for each sample.
1. A video decoding method performed by a decoding device, the method comprising:
deriving a reference picture list 0 (L 0 ) and a reference picture list 1 (L 1 );
deriving two motion vectors (MVs) for a current block, wherein the two MVs include an MVL 0 for the L 0 and an MVL 1 for the L 1 ;
obtaining bi-prediction optical-flow (BIO) usage flag from a picture parameter set (PPS) or a sequence parameter set (SPS) signaled from a bitstream;
determining whether to apply BIO prediction that derives a sample-unit MV in the current block based on the MVL 0 and the MVL 1 ;
deriving the sample-unit MV of the current block based on the MVL 0 and the MVL 1 based on a determination to apply the BIO prediction; and
deriving a prediction sample based on the sample-unit MV,
wherein the determining of whether to apply the BIO prediction comprises:
deriving a first time distance between the current picture and a first reference picture associated with the MVL 0 among the reference pictures included in the L 0 ;
deriving a second time distance between the current picture and a second reference picture associated with the MVL 1 among the reference pictures included in the L 1 ; and
determining, based on the BIO usage flag, whether to apply the BIO prediction based on the first time distance and the second time distance,
wherein the first time distance is a difference between a picture order count (POC) value of the current picture and a POC value of the first reference picture, and wherein the second time distance is a difference between a POC value of the second reference picture and the POC value of the current picture.
2. The method of claim 1 , wherein the sample-unit MV is derived based on the following equation:
Δ I [ i,j ]= I (0) [ i,j ]− I (1) [ i,j ]+ v x [ i,j ]( I x (0) [ i,j ]+ I x (1) [ i,j ])+ v y [ i,j ]( I y (0) [ i,j ]+ I y (1) [ i,j ])
where [i,j] is a coordinate of a target sample, Δ I[i, j] is a difference between a value of a first reference sample indicated by the MVL 0 in the first reference picture in the L 0 and a value of a second reference sample indicated by the MVL 1 in the second reference picture in the L 1 , I (0) [i,j] is a value of an L 0 reference sample at a position [i,j] in the first reference picture in the L 0 , I (1) [i,j] is a value of an L 1 reference sample at a position [i,j] in the second reference picture in the L 1 , I x (0) [i,j] and I y (0) [i,j] are respectively x-axis and y-axis derivative values at a position [i,j] of the first reference picture, I x (1) [i,j] and I y (1) [i,j] are respectively x-axis and y-axis derivative values at a position [i,j] of the second reference picture, v x [i,j] and v y [i,j] are respectively an x-component and y-component of the sample-unit MV for the target sample, and v x [i,j] and v y [i,j] have values which allow |Δ I[i, j] 2 to have a minimum value.
3. The method of claim 1 , wherein it is determined not to apply the BIO prediction if a value obtained by multiplying the x-component of the MVL 0 by the second time distance and a value obtained by multiplying the x-component of the MVL 1 by the first time distance have the same magnitude and different signs, and if a value obtained by multiplying the y-component of the MVL 0 by the second time distance and a value obtained by multiplying the y-component of the MVL 1 by the first time distance have the same magnitude and different signs, and wherein prediction samples of the current block are derived based on the MVL 0 and the MVL 1 if it is determined not to apply the BIO prediction.
4. The method of claim 1 , wherein it is determined not to apply the BIO prediction if the first time distance and the second time distance have the same magnitude, an x-component of the MVL 0 and an x-component of the MVL 1 have the same magnitude and different signs, and if a y-component of the MVL 0 and a y-component of the MVL 1 have the same magnitude and different signs, and wherein prediction samples of the current block are derived based on the MVL 0 and the MVL 1 if it is determined not to apply the BIO prediction.
5. The method of claim 1 , wherein it is determined not to apply the BIO prediction if a value obtained by multiplying the x-component of the MVL 0 by the second time distance and a value obtained by multiplying the x-component of the MVL 1 by the first time distance have the same magnitude and different signs, if a value obtained by multiplying the y-component of the MVL 0 by the second time distance and a value obtained by multiplying the y-component of the MVL 1 by the first time distance have the same magnitude and different signs, and if a size of the current block is 4×4, and wherein prediction samples of the current block are derived based on the MVL 0 and the MVL 1 if it is determined not to apply the BIO prediction.
6. The method of claim 1 , wherein it is determined not to apply the BIO prediction if the first time distance and the second time distance have the same magnitude, an x-component of the MVL 0 and an x-component of the MVL 1 have the same magnitude and different signs, if a y-component of the MVL 0 and a y-component of the MVL 1 have the same magnitude and different signs, and if a size of the current block is 4×4, and wherein prediction samples of the current block are derived based on the MVL 0 and the MVL 1 if it is determined not to apply the BIO prediction.
7. The method of claim 1 , wherein it is determined not to apply the BIO prediction if the merge mode is applied to the current block, and wherein prediction samples of the current block are derived based on the MVL 0 and the MVL 1 if it is determined not to apply the BIO prediction.
8. The method of claim 1 , further comprising obtaining a BIO usage flag through a bit-stream, wherein it is determined not to apply the BIO prediction if a value of the BIO usage flag is 0, and wherein prediction samples of the current block are derived based on the MVL 0 and the MVL 1 if it is determined not to apply the BIO prediction.
9. A decoding device for performing image decoding, comprising a prediction unit configured to derive a reference picture list 0 (L 0 ) and a reference picture list 1 (L 1 ), derive two motion vectors (MVs) for a current block, obtain bi-prediction optical-flow (BIO) usage flag from a picture parameter set (PPS) or a sequence parameter set (SPS) signaled from a bitstream, determine whether to apply bi prediction BIO prediction that derives a sample-unit MV in the current block based on the MVL 0 and the MVL 1 , derive the sample-unit MV of the current block based on the MVL 0 and the MVL 1 based on a determination to apply the BIO prediction, and derive a prediction sample based on the sample-unit MV,
wherein the two MVs include an MVL 0 for the L 0 and an MVL 1 for the L 1 ,
wherein the determining of whether to apply the BIO prediction comprises:
deriving a first time distance between the current picture and a first reference picture associated with the MVL 0 among the reference pictures included in the L 0 ;
deriving a second time distance between the current picture and a second reference picture associated with the MVL 1 among the reference pictures included in the L 1 ; and
determining, based on the BIO usage flag, whether to apply the BIO prediction based on the first time distance and the second time distance,
wherein the first time distance is a difference between a picture order count (POC) value of the current picture and a POC value of the first reference picture, and wherein the second time distance is a difference between a POC value of the second reference picture and the POC value of the current picture.
10. The decoding device of claim 9 , wherein the sample-unit MV is derived based on the following equation:
Δ I [ i,j ]= I (0) [ i,j ]− I (1) [ i,j ]+ v x [ i,j ]( I x (0) [ i,j ]+ I x (1) [ i,j ])+ v y [ i,j ]( I y (0) [ i,j ]+ I y (1) [ i,j ])
where [i,j] is a coordinate of a target sample, Δ I[i, j] is a difference between a value of a first reference sample indicated by the MVL 0 in the first reference picture in the L 0 and a value of a second reference sample indicated by the MVL 1 in the second reference picture in the L 1 , I (0) [i,j] is a value of an L 0 reference sample at a position [i,j] in the first reference picture in the L 0 , I (1) [i,j] is a value of an L 1 reference sample at a position [i,j] in the second reference picture in the L 1 , L x (0) [i,j] and I y (0) [i,j] are respectively x-axis and y-axis derivative values at a position [i,j] of the first reference picture, I x (1) [i,j] and I y (1) [i,j] are respectively x-axis and y-axis derivative values at a position [i,j] of the second reference picture, v x [i,j] and v y [i,j] are respectively an x-component and y-component of the sample-unit MV for the target sample, and v x [i,j] and v y [i,j] have values which allow |Δ I[i, j]| 2 to have a minimum value.
11. The decoding device of claim 9 , wherein the predictor is configured to: determine not to apply the BIO prediction if a value obtained by multiplying the x-component of the MVL 0 by the second time distance and a value obtained by multiplying the x-component of the MVL 1 by the first time distance have the same magnitude and different signs, and if a value obtained by multiplying the y-component of the MVL 0 by the second time distance and a value obtained by multiplying the y-component of the MVL 1 by the first time distance have the same magnitude and different signs; and derive prediction samples of the current block based on the MVL 0 and the MVL 1 if it is determined not to apply the BIO prediction.
12. The decoding device of claim 9 , wherein the prediction unit is configured to: determine not to apply the BIO prediction if the first time distance and the second time distance have the same magnitude, an x-component of the MVL 0 and an x-component of the MVL 1 have the same magnitude and different signs, and if a y-component of the MVL 0 and a y-component of the MVL 1 have the same magnitude and different signs; and derive prediction samples of the current block based on the MVL 0 and the MVL 1 if it is determined not to apply the BIO prediction.
13. The decoding device of claim 9 , wherein the prediction unit is configured to: determine not to apply the BIO prediction if a value obtained by multiplying the x-component of the MVL 0 by the second time distance and a value obtained by multiplying the x-component of the MVL 1 by the first time distance have the same magnitude and different signs, if a value obtained by multiplying the y-component of the MVL 0 by the second time distance and a value obtained by multiplying the y-component of the MVL 1 by the first time distance have the same magnitude and different signs, and if a size of the current block is 4×4; and derive prediction samples of the current block based on the MVL 0 and the MVL 1 if it is determined not to apply the BIO prediction.