Cross-component sample offset
In a method for filtering in video decoding, a first combination of filter tap values is determined based on reconstructed sample values of reconstructed samples of a first color component within a filter support region. Based on a lookup table (LUT) indicating a mapping that associates offset values with different combinations of filter tap values, a first offset value of the offset values in the mapping that is associated with the first combination of filter tap values is determined. The first offset value on a to-be filtered sample of a second color component in the filter support region is applied to determine a filtered sample of the second color component.
1 . A method for filtering in video decoding, the method comprising:
determining a first combination of filter tap values based on reconstructed sample values of reconstructed samples of a first color component within a filter support region;
determining, based on a lookup table (LUT) that includes offset values and different combinations of filter tap values that are associated with the offset values, a first offset value of the offset values that is associated with the first combination of filter tap values; and
applying the first offset value on a to-be filtered sample of a second color component in the filter support region to determine a filtered sample of the second color component.
2 . The method of claim 1 , wherein
the reconstructed samples of the first color component within the filter support region include (i) a collocated sample located at a same relative position as the to-be filtered sample of the second color component and (ii) surrounding reconstructed samples that surround the collocated sample, and
the determining the first combination of filter tap values includes:
calculating delta values that are differences between the respective surrounding reconstructed samples and the collocated sample; and
quantizing the delta values based on a quantization step size N to determine the first combination of filter tap values.
3 . The method of claim 2 , wherein for each delta value, the quantized delta value is one of −1, 0, and 1 based on a comparison of the respective delta value and the quantization step size N.
4 . The method of claim 3 , wherein
N is a positive integer, and
for each delta value,
the quantized delta value is −1 when the respective delta value<−N;
the quantized delta value is 0 when −N≤the respective delta value≤N; and
the quantized delta value is 1 when the respective delta value>N.
5 . The method of claim 4 , wherein the LUT associates the offset values with a complete domain of combinations for the reconstructed samples of the first color component within the filter support region, and the LUT includes the complete domain of combinations.
6 . The method of claim 5 , wherein
the reconstructed samples of the first color component within the filter support region include (i) a collocated sample located at a same relative position as the to-be filtered sample of the second color component and (ii) four surrounding reconstructed samples that surround the collocated sample in four respective directions, and
the complete domain of combinations in the LUT are 81 different combinations.
7 . The method of claim 1 , wherein the filter support region of the to-be filtered sample includes a sample position collocated with the to-be filtered sample of the second color component and at least two surrounding sample positions that surround the collocated sample position.
8 . The method of claim 1 , further comprising:
adding the first offset value to the to-be filtered sample of the second color component to determine an intermediate filtered sample; and
clipping the intermediate filtered sample based on a maximum value and a minimum value to determine the filtered sample value of the second color component.
9 . A method for filtering in video encoding, the method comprising:
determining a first combination of filter tap values based on reconstructed sample values of reconstructed samples of a first color component within a filter support region;
determining, based on a lookup table (LUT) that includes offset values and different combinations of filter tap values that are associated with the offset values, a first offset value of the offset values that is associated with the first combination of filter tap values; and
applying the first offset value on a to-be filtered sample of a second color component in the filter support region to determine a filtered sample of the second color component.
10 . The method of claim 9 , wherein
the reconstructed samples of the first color component within the filter support region include (i) a collocated sample located at a same relative position as the to-be filtered sample of the second color component and (ii) surrounding reconstructed samples that surround the collocated sample, and
the determining the first combination of filter tap values includes:
calculating delta values that are differences between the respective surrounding reconstructed samples and the collocated sample; and
quantizing the delta values based on a quantization step size N to determine the first combination of filter tap values.
11 . The method of claim 10 , wherein for each delta value, the quantized delta value is one of −1, 0, and 1 based on a comparison of the respective delta value and the quantization step size N.
12 . The method of claim 11 , wherein
N is a positive integer, and
for each delta value,
the quantized delta value is −1 when the respective delta value<−N;
the quantized delta value is 0 when −N≤the respective delta value≤N; and
the quantized delta value is 1 when the respective delta value>N.
13 . The method of claim 12 , wherein the LUT associates the offset values with a complete domain of combinations for the reconstructed samples of the first color component within the filter support region, and the LUT includes the complete domain of combinations.
14 . The method of claim 13 , wherein
the reconstructed samples of the first color component within the filter support region include (i) a collocated sample located at a same relative position as the to-be filtered sample of the second color component and (ii) four surrounding reconstructed samples that surround the collocated sample in four respective directions, and
the complete domain of combinations in the LUT are 81 different combinations.
15 . The method of claim 9 , wherein the filter support region of the to-be filtered sample includes a sample position collocated with the to-be filtered sample of the second color component and at least two surrounding sample positions that surround the collocated sample position.
16 . The method of claim 9 , further comprising:
adding the first offset value to the to-be filtered sample of the second color component to determine an intermediate filtered sample; and
clipping the intermediate filtered sample based on a maximum value and a minimum value to determine the filtered sample value of the second color component.
17 . A non-transitory computer-readable storage medium storing instructions which when executed by a processor cause the processor to perform a method of encoding a bitstream comprising:
determining a first combination of filter tap values based on reconstructed sample values of reconstructed samples of a first color component within a filter support region;
determining, based on a lookup table (LUT) that includes offset values and different combinations of filter tap values that are associated with the offset values, a first offset value of the offset values that is associated with the first combination of filter tap values;
applying the first offset value on a to-be filtered sample of a second color component in the filter support region to determine a filtered sample of the second color component; and
encoding and transmitting the bitstream indicating the filtered sample.
18 . The method of claim 17 , wherein
the reconstructed samples of the first color component within the filter support region include (i) a collocated sample located at a same relative position as the to-be filtered sample of the second color component and (ii) surrounding reconstructed samples that surround the collocated sample, and
the determining the first combination of filter tap values includes:
calculating delta values that are differences between the respective surrounding reconstructed samples and the collocated sample; and
quantizing the delta values based on a quantization step size N to determine the first combination of filter tap values.
19 . The method of claim 18 , wherein for each delta value, the quantized delta value is one of −1, 0, and 1 based on a comparison of the respective delta value and the quantization step size N.
20 . The method of claim 19 , wherein
N is a positive integer, and
for each delta value,
the quantized delta value is −1 when the respective delta value<−N;
the quantized delta value is 0 when −N≤the respective delta value≤N; and
the quantized delta value is 1 when the respective delta value>N.