Local illumination compensation for video encoding or decoding
In at least one embodiment, a method and apparatus for encoding/decoding a picture comprising predicting at least one block, wherein the predicting comprises performing motion compensation and local illumination compensation based on a reference block, the local illumination compensation including applying a linear model based on sums of absolute differences of neighboring reconstructed samples and corresponding reference samples of the reference block, wherein the neighboring reconstructed samples and corresponding reference samples of the reference block are co-located according to an L-shape substantially adjacent to the block to be predicted, the L-shape comprising a row of pixels located to the top side of the predicted block and a column of pixels located to the left side of the predicted block, the co-location being determined according a motion vector of the predicted block.
1. A video encoding method comprising predicting picture data for at least one block in a picture, wherein the predicting comprises performing motion compensation and local illumination compensation based on a reference block, the local illumination compensation comprising applying a linear model defined by at least a first parameter determined as a ratio between a sum of absolute first differences, each first difference being between a neighboring reconstructed sample and an average value of the neighboring reconstructed samples in a first L-shape and a sum of absolute second differences, each second difference being between a reference sample of the reference block and an average value of the reference samples in a second L-shape, said first L-shape being adjacent to the at least one block and said second L-shape being adjacent to the reference block.
2. The method of claim 1 , wherein the linear model is further defined by a second parameter, the first and second parameters being computed by:
a
=
(
∑
abs
(
cur
(
r
)
-
∑
cur
(
r
)
N
)
∑
abs
(
ref
(
s
)
-
∑
ref
(
s
)
N
)
)
b
=
∑
cur
(
r
)
N
-
a
×
∑
ref
(
s
)
N
wherein a is the first parameter, b is the second parameter, cur(r) is a neighboring reconstructed sample in a current block, ref(s) is a reference sample built with motion compensation translated by a motion vector my from the reference block, and s=r+mv.
3. The method of claim 1 , wherein the first parameter of the linear model is computed by:
a
=
(
∑
abs
(
cur
(
r
)
-
∑
cur
(
r
)
N
)
+
corr
∑
abs
(
ref
(
s
)
-
∑
ref
(
s
)
N
)
+
corr
)
where corr is a regularization term.
4. The method of claim 3 , wherein the corr regularization term is determined by:
corr
=
(
∑
abs
(
cur
(
r
)
-
∑
cur
(
r
)
N
)
)
⪢
reg_shift
where reg_shift is a shift value.
5. The method of claim 4 , wherein the reg_shift value is equal to 7.
6. An apparatus, comprising an encoder for encoding picture data for at least one block in a picture or video, wherein the encoder is configured to predict at least one block, wherein the predicting comprises performing motion compensation and local illumination compensation based on a reference block, the local illumination compensation comprising applying a linear model defined by at least a first parameter determined as a ratio between a sum of absolute first differences, each first difference being between a neighboring reconstructed sample and an average value of the neighboring reconstructed samples in a first L-shape and a sum of absolute second differences, each second difference being between a reference sample of the reference block and an average value of the reference samples in a second L-shape, said first L-shape being adjacent to the at least one block and said second L-shape being adjacent to the reference block.
7. The apparatus of claim 6 , wherein the linear model is further defined by a second parameter, the first and second parameters being computed by:
a
=
(
∑
abs
(
cur
(
r
)
-
∑
cur
(
r
)
N
)
∑
abs
(
ref
(
s
)
-
∑
ref
(
s
)
N
)
)
b
=
∑
cur
(
r
)
N
-
a
×
∑
ref
(
s
)
N
wherein a is the first parameter, b is the second parameter, cur(r) is a neighboring reconstructed sample in a current block, ref(s) is a reference sample built with motion compensation translated by a motion vector my from the reference block, and s=r+mv.
8. The apparatus of claim 6 , wherein the first parameter of the linear model is computed by:
a
=
(
∑
abs
(
cur
(
r
)
-
∑
cur
(
r
)
N
)
+
corr
∑
abs
(
ref
(
s
)
-
∑
ref
(
s
)
N
)
+
corr
)
where corr is a regularization term.
9. The apparatus of claim 8 , wherein the corr regularization term is determined by:
corr
=
(
∑
abs
(
cur
(
r
)
-
∑
cur
(
r
)
N
)
)
⪢
reg_shift
where reg_shift is a shift value.
10. The apparatus of claim 9 , wherein the reg_shift value is equal to 7.
11. A non-transitory computer readable medium comprising program code instructions which, when the program is executed by a computer, cause the computer to carry out a video decoding method comprising predicting picture data for at least one block in a picture, wherein the predicting comprises performing motion compensation and local illumination compensation based on a reference block, the local illumination compensation comprising applying a linear model defined by at least a first parameter determined as a ratio between a sum of absolute first differences, each first difference being between a neighboring reconstructed sample and an average value of the neighboring reconstructed samples in a first L-shape and a sum of absolute second differences, each second difference being between a reference sample of the reference block and an average value of the reference samples in a second L-shape, said first L-shape being adjacent to the at least one block and said second L-shape being adjacent to the reference block.
12. A video decoding method comprising predicting picture data for at least one block in a picture, wherein the predicting comprises performing motion compensation and local illumination compensation based on a reference block, the local illumination compensation comprising applying a linear model defined by at least a first parameter determined as a ratio between a sum of absolute first differences, each first difference being between a neighboring reconstructed sample and an average value of the neighboring reconstructed samples in a first L-shape and a sum of absolute second differences, each second difference being between a reference sample of the reference block and an average value of the reference samples in a second L-shape, said first L-shape being adjacent to the at least one block and said second L-shape being adjacent to the reference block.
13. The method of claim 12 , wherein the linear model is further defined by a second parameter, the first and second parameters being computed by:
a
=
(
∑
abs
(
cur
(
r
)
-
∑
cur
(
r
)
N
)
∑
abs
(
ref
(
s
)
-
∑
ref
(
s
)
N
)
)
b
=
∑
cur
(
r
)
N
-
a
×
∑
ref
(
s
)
N
wherein a is the first parameter, b is the second parameter, cur(r) is a neighboring reconstructed sample in a current block, ref(s) is a reference sample built with motion compensation translated by a motion vector mv from the reference block, and s=r+mv.
14. The method of claim 12 , wherein the first parameter of the linear model is computed by:
a
=
(
∑
abs
(
cur
(
r
)
-
∑
cur
(
r
)
N
)
+
corr
∑
abs
(
ref
(
s
)
-
∑
ref
(
s
)
N
)
+
corr
)
where corr is a regularization term.
15. The method of claim 14 , wherein the corr regularization term is determined by:
corr
=
-
(
∑
abs
(
cur
(
r
)
-
∑
cur
(
r
)
N
)
)
⪢
reg_shift
where reg_shift is a shift value.
16. The method of claim 15 , wherein the reg_shift value is equal to 7.
17. An apparatus, comprising an decoder for decoding picture data for at least one block in a picture or video wherein the decoder is configured to predict at least one block, wherein the predicting comprises performing motion compensation and local illumination compensation based on a reference block, the local illumination compensation comprising applying a linear model defined by at least a first parameter determined as a ratio between a sum of absolute first differences, each first difference being between a neighboring reconstructed sample and an average value of the neighboring reconstructed samples in a first L-shape and a sum of absolute second differences, each second difference being between a reference sample of the reference block and an average value of the reference samples in a second L-shape, said first L-shape being adjacent to the at least one block and said second L-shape being adjacent to the reference block.
18. The apparatus of claim 17 , wherein the linear model is further defined by a second parameter, the first and second parameters being computed by:
a
=
(
∑
abs
(
cur
(
r
)
-
∑
cur
(
r
)
N
)
∑
abs
(
ref
(
s
)
-
∑
ref
(
s
)
N
)
)
b
=
∑
cur
(
r
)
N
-
a
×
∑
ref
(
s
)
N
wherein a is the first parameter, b is the second parameter, cur(r) is a neighboring reconstructed sample in a current block, ref(s) is a reference sample built with motion compensation translated by a motion vector my from the reference block, and s=r+mv.
19. The apparatus of claim 17 , wherein the first parameter of the linear model is computed by:
a
=
(
∑
abs
(
cur
(
r
)
-
∑
cur
(
r
)
N
)
+
corr
∑
abs
(
ref
(
s
)
-
∑
ref
(
s
)
N
)
+
corr
)
where corr is a regularization term.
20. The apparatus of claim 19 , wherein the corr regularization term is determined by:
corr
=
-
(
∑
abs
(
cur
(
r
)
-
∑
cur
(
r
)
N
)
)
⪢
reg_shift
where reg_shift is a shift value.