Sparsity-Based De-Artifact Filtering for Video Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video encoding and decoding technologies, such as the MPEG-4 AVC Standard, struggle to effectively remove artifacts like ringing, edge distortion, and texture corruption caused by coarse quantization, as the deblocking filter is inadequate for handling these issues inside blocks and is not suited for image singularities like edges or textures.
Innovation Solution
A sparsity-based de-artifact filtering method is introduced, which groups image regions, transforms them, applies adaptive de-artifact filtering, and then inverse transforms the filtered regions to restore them to their original positions, using a de-artifacting filter that can replace or follow the deblocking filter, and is configured to operate in the transform domain with adaptive filtering.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If a deblocking filter is used to reduce blocky artifacts, then block boundary artifacts are improved, but artifacts inside blocks (ringing, edge distortion, texture corruption) remain unaffected
Solution Approach 1:
The patent segments the image into multiple overlapping regions or patches, and applies different filtering operations to different segments based on their local characteristics. This allows the filter to address both block boundary artifacts and internal artifacts (ringing, edge distortion, texture corruption) by adapting to each region's specific needs.
Solution Approach 2:
The patent employs dynamic filtering where the filtering strength and type are adaptively adjusted based on local image characteristics such as edge detection results, variance measurements, and artifact detection. This dynamic adaptation enables the filter to effectively handle different artifact types in different regions rather than applying a static filtering approach.
2Productivity
If coarse quantization is applied to reduce bitrate, then compression efficiency is improved, but visual quality deteriorates due to quantization artifacts
Solution Approach 1:
The patent introduces a de-artifacting filter as an intermediary processing stage between quantization and final output. This filter acts as a mediator that removes quantization artifacts (ringing, edge distortion, texture corruption) while preserving the compressed representation, thereby decoupling the trade-off between compression efficiency and visual quality.
Solution Approach 2:
The patent changes the parameter space by operating in the transform domain (e.g., DCT or wavelet domain) rather than purely in the spatial domain. By transforming the image, applying adaptive filtering in the transform domain, and then inverse transforming, the system can more effectively remove quantization artifacts while maintaining compression efficiency.
3Object-affected harmful factors
If low-pass filters are applied to block edges, then blocky artifacts are reduced, but edge sharpness and texture details may be degraded
Solution Approach 1:
The patent applies local quality by using different filtering strengths and types for different regions. Edge regions detect ed ed edges are identified and protected from excessive smoothing, while flat regions receive stronger filtering. This local adaptation preserves edge sharpness and texture details while still removing blocky artifacts where appropriate.
Solution Approach 2:
The patent uses dynamic filtering where the filter strength is adaptively adjusted based on local image characteristics such as gradient magnitude, variance, and edge detection results. This dynamic approach prevents over-smoothing of edges and textures while effectively reducing blocky artifacts in appropriate regions.
Data Source
AI summary
Methods and apparatus are provided for sparsity-based de-artifact filtering for video encoding and decoding. An apparatus includes an encoder (400) for encoding at least a portion of an image by grouping regions within the portion based on a grouping metric, transforming the grouped regions, adaptively performing de-artifact filtering on the transformed regions using a de-artifacting filter (413) included in the encoder, inverse transforming the de-artifacted regions to create replacement regions, and restoring the replacement regions to positions with the image from which the regions were taken prior to the grouping.