Adaptive Loop Filtering Using Residual Side Information in Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face challenges in efficiently managing bandwidth demands for digital video due to the increasing number of connected devices, particularly in handling high-resolution videos with complex chroma subsampling and filtering processes, which can lead to artifacts and inefficiencies in compression.
Innovation Solution
The implementation of an adaptive loop filter (ALF) that utilizes residual samples as side information for video coding, enabling conversion between visual media data and bitstreams, and optimizing filtering processes at the picture level to minimize mean square errors and adapt to different regions within a picture.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If adaptive loop filter (ALF) uses residual samples as side information, then coding efficiency is improved and artifacts are reduced, but device complexity increases
Solution Approach 1:
The patent changes the input parameter of the ALF from conventional filtered samples to residual samples. This parameter change enables the filter to operate on different data characteristics, improving coding efficiency by better utilizing the residual information while maintaining manageable complexity through the mathematical properties of residual samples.
Solution Approach 2:
The patent introduces residual samples as an intermediary element between the prediction process and the loop filter. These residual samples serve as side information that mediates the filtering process, allowing the ALF to compensate for prediction errors more effectively without directly processing the complex original and predicted sample differences.
2Device complexity
If conventional filtering processes are used, then processing is simpler, but artifacts appear and compression efficiency decreases
Solution Approach 1:
The patent fundamentally changes the input parameter of the loop filter from conventional reconstructed samples to residual samples. This parameter transformation allows the filter to target prediction errors directly, reducing artifacts caused by conventional filtering limitations while maintaining reasonable processing complexity through efficient residual sample utilization.
Solution Approach 2:
The patent implements a feedback mechanism where residual samples (which represent the difference between original and predicted values) are fed back into the ALF process. This feedback loop enables the filter to continuously compensate for prediction inaccuracies, reducing artifacts by using the residual information to adjust and improve the reconstructed output.
Data Source
AI summary
A mechanism for processing video data is disclosed. The mechanism includes determining to employ an adaptive loop filter (ALF) that receives a residual sample of a current picture as side information used as an input. A conversion is performed between a visual media data and a bitstream based on the ALF.