Adaptive non-linear mapping for sample offset
A method for in-loop sample offset filtering in a video decoder is disclosed. The method includes obtaining at least one statistical property associated with reconstructed samples of at least a first color component in a current reconstructed data block of a video stream, selecting a target sample offset filter among a plurality of sample offset filters based on the at least one statistical property, the target sample offset filter comprising a nonlinear mapping between sample delta measures and sample offset values, and filtering a current sample in a second color component of the current reconstructed data block using the target sample offset filter and reference samples in a third color component of the current reconstructed data block to generate a filtered reconstructed sample of the current sample.
1. A method for encoding a video, comprising:
determining two or more alternative sample offset filters with a same filtering tap pattern, the two or more alternative sample offset filters corresponding to two or more alternative lookup tables, the lookup tables all comprising a same plurality of possible sample delta measures as lookup keys, the same plurality of possible sample delta measures being mapped to different pluralities of offset values in the two or more alternative look up tables, each of the sample delta measures comprising one possible combination of pixel value differences in a quantized pixel value space between a plurality of reference tap positions relative to a filter tap position in the same filtering tap pattern;
obtaining at least one statistical property associated with reconstructed samples of at least a first color component in a data block of the video;
selecting a target sample offset filter among the two or more alternative sample offset filters having the same plurality of sample delta measures mapped to different pluralities of offset values based on the at least one statistical property for loop filtering of the reconstructed samples; and
including a signaling element in an encoded bitstream of the video to identify the target sample offset filter among the two or more alternative sample offset filters for the reconstructed samples.
2. The method of claim 1 , further comprising:
filtering the reconstructed samples in a second color component of the data block to generate filtered reconstructed samples in the second color component using the target sample offset filter; and
storing the filtered reconstructed samples in the second color component in a current picture buffer or a reference picture buffer as reference samples for encoding other inter-predicted or intra-predicted samples of the video.
3. The method of claim 2 , wherein filtering each of the reconstructed samples in the second color component of the data block using the target sample offset filter comprises:
selecting a sample offset from the plurality of offset values of the target sample offset filter based on a calculated combination of sample delta measures of a filter sample and corresponding reference samples at the plurality of reference tap positions of the filter sample in a third color component of the data block as reconstructed; and
applying the selected sample offset to the each of the reconstructed samples in the second color component to generate the filtered reconstructed samples in the second color component.
4. The method of claim 3 , wherein the first color component is a same color component as the third color component.
5. The method of claim 3 , wherein first color component is a same color component as the second color component.
6. The method of claim 3 , wherein the second color component is a different color component as the third color component.
7. The method of claim 3 , wherein the second color component is a same color component as the third color component.
8. The method of claim 3 , wherein the at least one statistical property comprises edge information of the data block.
9. The method of claim 8 , wherein:
the edge information of the data block comprises an edge direction derived in a Constrained Directional Enhanced Filtering (CDEF) process; and
the two or more alternative sample offset filters comprise N sample offset filters corresponding to N CDEF edge directions, N being an integer between 1 and 8 inclusive.
10. The method of claim 9 , wherein the two or more alternative sample offset filters are signaled at a frame level in high level syntax (HLS) in the encoded bitstream.
11. The method of claim 3 , wherein filtering each of the reconstructed samples in the second color component of the data block using the target sample offset filter comprises:
determining a first location of each of the reconstructed samples of the second color component;
identifying the filter sample of a third color component at the first location and the reference samples at the plurality of reference tap positions relative to the first location in the third color component;
determining the calculated combination of sample delta measures between the reference samples and the filter sample of the third color component of the data block;
extracting the sample offset from the plurality of offset values of the target sample offset filter based on the calculated combination of sample delta measures; and
applying the sample offset to each of the reconstructed samples so as to generate the filtered reconstructed samples of the reconstructed samples in the second color component.
12. The method of claim 2 , wherein the at least one statistical property comprises a smoothness measure of the data block.
13. The method of claim 12 , wherein:
the smoothness measure of the data block is mapped to one of M predefined smoothness levels characterized by M-1 smoothness level thresholds;
the two or more alternative sample offset filters comprise M sample offset filters corresponding to the M predefined smoothness levels; and
the target sample offset filter is selected from the M sample offset filters according the one of the M predefined smoothness levels mapped to the smoothness measure.
14. The method of claim 2 , wherein the at least one statistical property comprises a coding information of the data block.
15. The method of claim 14 , wherein
coding information comprises a current prediction mode of the data block;
the two or more alternative sample offset filters correspond to different prediction modes; and
the target sample offset filter is selected from the two or more alternative sample offset filters according to the current prediction mode of the data block.
16. The method of claim 15 , wherein the different prediction modes comprise at least one of intra DC mode, intra Planar mode, intra PAETH mode, intra SMOOTH mode, intra recursive filtering mode, and inter SKIP mode.
17. The method of claim 2 , wherein each of the two or more alternative sample offset filters is associated with a set of filter coefficients, a number of the plurality of reference tap positions, and the plurality of reference tap positions.
18. The method of claim 2 , wherein:
the two or more alternative sample offset filters are predetermined;
the two or more alternative sample offset filters and an index of the selected target sample offset filter among the two or more alternative sample offset filters are signaled at a sequence level, a picture level, or coding tree unit level in the encoded bitstream.
19. A video decoder comprising a processing circuitry configured to:
determine two or more alternative sample offset filters with a same filtering tap pattern for providing two or more alternative mappings each comprising a plurality of offset values nonlinearly mapped to a plurality of possible sample delta measures, each of the sample delta measures comprising one possible combination of pixel value differences in a quantized pixel value space between a plurality of reference tap positions relative to a filter tap position in the filtering tap pattern;
obtain at least one statistical property associated with reconstructed samples of at least a first color component in a data block of a video;
select a target sample offset filter among the two or more alternative sample offset filters having the sample plurality of sample delta measures mapped to different pluralities of offset values based on the at least one statistical property for loop filtering of the reconstructed samples; and
include a signaling element in an encoded bitstream of the video to identify the target sample offset filter among the two or more alternative sample offset filters for the reconstructed samples.
20. A non-transitory computer readable medium for storing computer instructions, the computer instructions, when executed by a processor in a video encoder, causes the video encoder to:
determine two or more alternative sample offset filters with a same filtering tap pattern, the two or more alternative sample offset filters corresponding to two or more alternative lookup tables, the lookup tables all comprising a same plurality of possible sample delta measures as lookup keys, the same plurality of possible sample delta measures being mapped to different pluralities of offset values in the two or more alternative look up tables, each of the sample delta measures comprising one possible combination of pixel value differences in a quantized pixel value space between a plurality of reference tap positions relative to a filter tap position in the same filtering tap pattern;
obtain at least one statistical property associated with reconstructed samples of at least a first color component in a data block of a video;
select a target sample offset filter among the two or more alternative sample offset filters having the same plurality of sample delta measures mapped to different pluralities of offset values based on the at least one statistical property for loop filtering of the reconstructed samples; and
include a signaling element in an encoded bitstream of the video to identify the target sample offset filter among the two or more alternative sample offset filters for the reconstructed samples.