Sub-Shape Adaptive Loop Filtering for Lower-Complexity Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding systems, such as VVC, face inefficiencies in adaptive loop filtering due to complex block classification and filter coefficient signaling, particularly in handling chroma components, leading to increased computational complexity and potential amplification of artifacts.
Innovation Solution
The introduction of sub-shape Adaptive Loop Filters (ALFs) that selectively disable certain taps or types of differences in the ALF process, reducing computational complexity and improving filtering efficiency by adapting filter shapes based on specific conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If full-shape ALF with all taps enabled is used, then filtering quality is improved, but computational complexity increases
Solution Approach 1:
The patent segments the full-shape ALF into multiple sub-shapes by selectively disabling certain taps. This creates a hierarchy of filter configurations (full-shape, sub-shape 1, sub-shape 2, etc.) where each sub-shape disables specific taps based on their contribution to filtering quality. The encoder can select the appropriate sub-shape level to balance quality and complexity, rather than always using the computationally intensive full-shape filter.
Solution Approach 2:
The patent applies partial action by selectively enabling only the necessary taps required for adequate filtering performance. Instead of applying all taps (excessive action), the system determines the minimum sufficient filter configuration by disabling taps that contribute minimally to quality improvement, thereby reducing computational complexity while maintaining acceptable filtering quality.
2Adaptability or versatility
If more filter coefficients are signaled, then filtering adaptability is improved, but bitstream overhead increases
Solution Approach 1:
The patent segments the filter coefficient signaling into groups corresponding to different tap types. By disabling certain taps in sub-shape configurations, the system only needs to signal coefficients for the remaining active taps. This segmentation allows selective signaling of only necessary coefficients, reducing bitstream overhead while preserving adaptability for the enabled taps.
Solution Approach 2:
Different tap types contribute differently to filtering quality for different video content characteristics. The patent applies local quality by selectively disabling taps based on their specific contribution patterns, allowing the system to reduce coefficient signaling for taps that provide minimal local quality improvement in certain contexts, thereby reducing overall bitstream overhead.
3Manufacturing precision
If chroma component filtering is applied, then video quality is improved, but artifact amplification occurs
Solution Approach 1:
The patent applies local quality by selectively disabling certain taps in chroma component filtering based on their contribution to quality versus artifact generation. Different tap types affect chroma quality differently, and the system identifies and disables specific taps that tend to amplify artifacts while preserving taps that improve quality, achieving localized optimization for chroma filtering.
Solution Approach 2:
The patent converts the potentially harmful effect of certain taps (which amplify artifacts) into a benefit by selectively disabling them. What would normally be a harmful artifact-amplifying tap is identified and disabled, transforming the filtering process to eliminate the harmful effect while preserving the beneficial quality-improving taps.
Data Source
AI summary
A method and apparatus for video coding using sub-shape ALF (Adaptive Loop Filter). According to the method, reconstructed pixels associated with a current block are received. A full-shape ALF is determined. At least one sub-shape ALF is derived from the full-shape ALF by setting one or more taps of the full-shape ALF to 0. A current filtered output is derived by applying a target ALF to the current block, wherein the target ALF is selected from ALF candidates comprising said at least one sub-shape ALF. Filtered-reconstructed pixels are then provided, wherein the filtered-reconstructed pixels comprise the current filtered output.