Guided Video Filtering for Post-Decoding Ringing Removal
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies struggle to effectively remove ringing artifacts caused by transform and quantization processes, which are not adequately addressed by current in-loop filters.
Innovation Solution
The application of a guided filter at various stages of the video processing pipeline, including post-decoding, to correct ringing artifacts by classifying video units into groups based on statistical information and applying tailored filtering parameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If guided filter is applied at post decoding stage, then ringing artifacts are significantly reduced, but processing time and complexity increase
Solution Approach 1:
The guided filter is applied at the post decoding stage, performing artifact removal before final video output. This preliminary action ensures that ringing artifacts are eliminated early in the processing pipeline, improving perceived quality without requiring re-processing later.
Solution Approach 2:
The filter classifies video units into different groups based on statistical information (variance, gradient magnitude) and applies different filtering parameters to different groups. This local quality approach ensures that regions with strong edges use different parameters than flat regions, optimizing artifact removal while preserving important video features.
2Measurement precision
If guided filter classifies video units into multiple groups with different parameters, then filtering accuracy improves, but computational complexity increases
Solution Approach 1:
The patent implements classification of video units into multiple groups based on local statistical properties (variance, gradient magnitude). Each group receives tailored filtering parameters, achieving high filtering accuracy by adapting to local content characteristics while maintaining manageable computational complexity through efficient classification thresholds.
Solution Approach 2:
Different filtering parameters (regularization weights, filter window sizes) are applied to different video unit groups based on their statistical characteristics. This parameter adaptation allows the filter to optimize performance for each local region type (flat, edge, texture) without requiring a completely different filtering approach for each case.
3Manufacturing precision
If guided filter is applied to all video units, then overall video quality improves, but processing overhead increases
Solution Approach 1:
Rather than applying uniform filtering to all video units, the patent classifies video units into different groups and applies filtering selectively with different parameters. This local quality approach improves overall video quality by adapting to local content while reducing processing overhead by avoiding unnecessary heavy filtering in regions where it is less needed.
Solution Approach 2:
The filter applies different levels of processing intensity to different video unit groups. Some groups receive stronger filtering (excessive action) while others receive lighter processing (partial action), optimizing the balance between quality improvement and processing efficiency by avoiding uniform maximum processing across all regions.
Data Source
AI summary
A mechanism for processing video data is disclosed. A conversion is performed between a visual media data including a plurality of video units and a bitstream. A guided filter is applied to samples in the video units. The guided filter is applied at a post decoding stage, such as in an in-loop filtering process.