Adaptive Loop Filter Switching via Projection-Slice Block Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies, such as AVS HPM-6, utilize region-based adaptive loop filters that require encoding multiple filters, increasing bitstream size, while newer standards like VVC use block-based adaptive loop filters that are gradient-based, which may not be optimal for coding efficiency.
Innovation Solution
Implementing a block-based adaptive loop filtering method for AVS3 HPM-6 using the projection-slice theorem to select filters at 4x4 block levels, reducing the number of filters encoded in the bitstream and improving coding efficiency without relying on gradient-based methods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple test patterns are used to accurately represent various image characteristics for ALF classification, then classification accuracy is improved, but the complexity of the deblocking filter determination process increases
Solution Approach 1:
The patent segments the image processing task by dividing the image into multiple blocks and applying different deblocking filters to different blocks based on their specific characteristics. This allows accurate classification without requiring a single complex filter to handle all cases, thereby resolving the contradiction between classification accuracy and process complexity.
Solution Approach 2:
The patent applies different deblocking filter strengths (first, second, or third filters) to different image blocks based on local characteristics such as block boundary strength and gradient values. This local adaptation improves classification accuracy while avoiding the need for a universally complex filtering process.
2Manufacturing precision
If complex deblocking filters are applied to all image blocks, then filtering quality is improved, but the computational complexity and processing time increase
Solution Approach 1:
The patent changes the filtering parameters (filter type and strength) based on the characteristics of each image block. By adjusting parameters locally rather than applying a fixed complex filter everywhere, the patent achieves high filtering quality while maintaining processing efficiency.
Solution Approach 2:
The patent applies strong filtering only where necessary (in blocks with high block boundary strength or specific gradient characteristics) and uses lighter filtering elsewhere. This partial application of complex filtering maintains quality where needed while preserving overall processing speed.
3Measurement precision
If ALF is applied to 4:2:0 chroma samples, then color accuracy is improved, but the risk of color artifacts and processing complexity increase
Solution Approach 1:
The patent applies ALF filtering to chroma samples locally based on the characteristics of each chroma block. By evaluating block-specific metrics and applying filtering only where appropriate, the patent improves color accuracy while minimizing the risk of color artifacts that could arise from uniform application.
Solution Approach 2:
The patent uses feedback from block characteristic analysis (such as gradient calculations and boundary strength measurements) to determine whether ALF should be applied to chroma samples. This feedback mechanism ensures color accuracy is improved only when beneficial, preventing color artifacts.
Data Source
Figure 1
Figure 2A
Figure 2B
AI summary
A method, computer program, and computer system is provided for video coding. Video data comprising sub-blocks corresponding to neighborhood data is received. AC projections are computed based on at least a portion of the neighborhood data, and AC projections in the same direction are aggregated. One or more AC energy indices and one or more directionality indices are computed based on the aggregated AC projections. A class index is computed based on the computed AC energy and directionality indices. The video data is decoded based on the computed class index.