Cross-component Sample Adaptive Offset

1 . A method for Cross-Component Sample Adaptive Offset (CCSAO) process, comprising:

determining a first rate-distortion (RD) cost for a block using a first classifier, wherein the first classifier has a first category number in a first value range, and wherein the first RD cost is the least among RD costs associated with classifiers having a category number in the first value range;

determining a second RD cost for the block using a second classifier, wherein the second classifier has a second category number in a second value range, and wherein the second RD cost is the least among RD costs associated with classifiers having a category number in the second value range; and

applying the first classifier as a classifier for the CCSAO process in response to determining the first RD cost is less than the second RD cost, wherein the first classifier classifies samples of a first component based at least in part on samples of a second component; and

repeating the determining an RD cost for the block using an additional classifier having a category number in a next value range in response to determining the RD cost is decreasing until a threshold value range is reached and applying a last used classifier as the classifier for the CCSAO process.

2 . The method of claim 1 , further comprising:

determining an initial RD cost for the block; and

determining that the first RD cost is less than the initial RD cost.

3 . The method of claim 1 , wherein the first value range and the second value range are adjacent ranges or overlapping ranges.

4 . The method of claim 1 , wherein the first classifier and the second classifier use EO (Edge Offset) or BO (Band Offset) or a combination thereof for classification.

5 . The method of claim 1 , wherein a classifier is skipped in response to determining that the classifier comprises a BO classifier in which the number of bands for Y component is less than the number of bands for U component or the number of bands for V component.

6 . An apparatus for Cross-Component Sample Adaptive Offset (CCSAO) process, comprising:

a memory; and

at least one processor coupled to the memory and configured to perform a method comprising:

determining a first rate-distortion (RD) cost for a block using a first classifier, wherein the first classifier has a first category number in a first value range, and wherein the first RD cost is the least among RD costs associated with classifiers having a category number in the first value range;

determining a second RD cost for the block using a second classifier, wherein the second classifier has a second category number in a second value range, and wherein the second RD cost is the least among RD costs associated with classifiers having a category number in the second value range; and

applying the first classifier as a classifier for the CCSAO process in response to determining the first RD cost is less than the second RD cost, wherein the first classifier classifies samples of a first component based at least in part on samples of a second component; and

repeating the determining an RD cost for the block using an additional classifier having a category number in a next value range in response to determining the RD cost is decreasing until a threshold value range is reached and applying a last used classifier as the classifier for the CCSAO process.

7 . The apparatus of claim 6 , wherein the method further comprises:

determining an initial RD cost for the block; and

determining that the first RD cost is less than the initial RD cost.

8 . The apparatus of claim 6 , wherein the first value range and the second value range are adjacent ranges or overlapping ranges.

9 . The apparatus of claim 6 , wherein the first classifier and second classifier use EO (Edge Offset) or BO (Band Offset) or a combination thereof for classification.

10 . The apparatus of claim 6 , wherein a classifier is skipped in response to determining that the classifier comprises a BO classifier in which the number of bands for Y component is less than the number of bands for U component or the number of bands for V component.

11 . A non-transitory computer readable storage medium storing a bitstream to be decoded by a method comprising:

determining a first rate-distortion (RD) cost for a block using a first classifier, wherein the first classifier has a first category number in a first value range, and wherein the first RD cost is the least among RD costs associated with classifiers having a category number in the first value range;

determining a second RD cost for the block using a second classifier, wherein the second classifier has a second category number in a second value range, and wherein the second RD cost is the least among RD costs associated with classifiers having a category number in the second value range; and

applying the first classifier as a classifier for a CCSAO process in response to determining the first RD cost is less than the second RD cost, wherein the first classifier classifies samples of a first component based at least in part on samples of a second component; and

repeating the determining an RD cost for the block using an additional classifier having a category number in a next value range in response to determining the RD cost is decreasing until a threshold value range is reached and applying a last used classifier as the classifier for the CCSAO process.

12 . The non-transitory computer readable storage medium of claim 11 , wherein the method further comprises:

determining an initial RD cost for the block; and

determining that the first RD cost is less than the initial RD cost.

13 . The non-transitory computer readable storage medium of claim 11 , wherein the first value range and the second value range are adjacent ranges or overlapping ranges.

14 . The non-transitory computer readable storage medium of claim 11 , wherein the first classifier and the second classifier use EO (Edge Offset) or BO (Band Offset) or a combination thereof for classification.

15 . The non-transitory computer readable storage medium of claim 11 , wherein a classifier is skipped in response to determining that the classifier comprises a BO classifier in which the number of bands for Y component is less than the number of bands for U component or the number of bands for V component.