CCSO Filtering With Fractional Taps for Video Reconstruction

The patent applies preliminary action by performing interpolation of the first color component to generate interpolated samples at fractional tap positions before applying the CCSO filter. This preliminary interpolation step enables the use of fractional tap positions without directly increasing filter complexity, as the interpolation is performed in advance using standard interpolation filters.

The patent introduces an intermediary approach by using a separate interpolation filter as a mediator between the reconstructed samples and the CCSO filter. The interpolation filter creates intermediate interpolated samples that bridge the gap between integer and fractional tap positions, allowing the CCSO filter to operate with fractional positions while maintaining manageable complexity through the separation of functions.

The patent applies segmentation by dividing the filtering process into two distinct segments: first, an interpolation step that generates interpolated samples at fractional positions; second, the CCSO filtering step that uses these interpolated samples. This segmentation allows the complex fractional tap positioning to be separated from the main filtering operation, reducing the complexity burden on any single component while achieving high overall quality.

The patent uses preliminary action by performing the interpolation operation before the CCSO filtering. This preliminary interpolation creates the necessary fractional position samples in advance, allowing the CCSO filter to focus solely on the offset filtering operation without having to handle the complexity of fractional position sampling itself, thus improving quality while managing complexity.

The patent applies local quality by adapting the number of interpolation taps based on the specific requirements of different regions or conditions in the video signal. Rather than uniformly applying high-tap interpolation throughout, the system can select appropriate tap counts (e.g., 2-tap, 4-tap, 6-tap, or 8-tap) based on local characteristics, reducing processing time in regions where fewer taps are sufficient while maintaining effective redundancy reduction where needed.

The patent employs partial action by using variable numbers of interpolation taps rather than always applying maximum interpolation. The system can use fewer taps (partial action) when the video content or position allows, reducing processing time, and reserve more extensive interpolation (excessive action relative to minimum needs) only when necessary to achieve adequate redundancy reduction, thus optimizing the balance between quality and processing time.