Adaptive Loop Filtering at Video Slice and Tile Boundaries
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding standards, such as HEVC, face challenges in efficiently performing loop filtering across slice and tile boundaries, leading to decreased visual quality due to the use of padded data or disabling of filters in these areas.
Innovation Solution
The implementation of adaptive loop filtering techniques using partial filters, such as asymmetric and symmetric partial filters, which skip unused filter coefficients across boundaries, allowing for renormalization to maintain filter effectiveness without requiring padded data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If loop filtering is performed across slice or tile boundaries using conventional methods, then filtering coverage is improved, but visual quality deteriorates due to use of padded data
Solution Approach 1:
The patent applies different filtering strategies to different regions: inside slice/tile boundaries conventional loop filtering is used, while at boundaries partial filters are applied that exclude padded data. This local differentiation maintains visual quality at boundaries while preserving adequate filtering coverage.
Solution Approach 2:
The filtering process is segmented into boundary regions and non-boundary regions. At slice/tile boundaries, the filter is divided to process only valid pixel data excluding padded areas, while interior regions use conventional full filters. This segmentation prevents quality degradation from padded data while maintaining overall filtering coverage.
2Manufacturing precision
If loop filtering is disabled at slice or tile boundaries to avoid padded data issues, then visual quality is maintained, but filtering coverage is reduced
Solution Approach 1:
Instead of completely disabling filtering at boundaries, the patent applies partial filters that perform filtering action on the portion of pixels that are valid (non-padded). This partial action maintains visual quality by excluding padded data while still providing filtering benefit to actual image content at boundary regions.
3Area of stationary object
If padded data is used to enable filtering at boundaries, then filtering coverage is improved, but computational efficiency deteriorates
Solution Approach 1:
The patent extracts and excludes padded data from the filtering process at slice/tile boundaries. By identifying and removing padded pixels from filter operations, it avoids the computational waste of processing invalid data while maintaining filtering coverage on valid boundary pixels through partial filter application.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The techniques of this disclosure apply to loop filtering across slice or tile boundaries in a video coding process. In one example, a method for performing loop filtering in a video coding process includes determining that pixels corresponding to filter coefficients of a filter mask for a loop filter are across a slice or tile boundary, removing filter coefficients corresponding to the pixels across the slice or tile boundary from the filter mask, renormalizing the filter mask without the removed filter coefficients, performing loop filtering using the renormalized filter mask