Adaptive Bilateral Filtering for Video Ringing Artifact Removal
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies struggle to completely remove ringing artifacts, particularly in chroma components, during the in-loop filtering process.
Innovation Solution
A bilateral filter is applied to video samples, with weights varying based on distance and intensity differences, to address ringing artifacts. The filter can be adaptive and is configured to operate on luma and chroma components differently, using statistical information and coded information to determine application and parameter settings.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If existing in-loop filtering is applied to video samples, then some artifacts are reduced, but ringing artifacts in chroma components cannot be completely removed
Solution Approach 1:
The patent applies a bilateral filter with dynamically adjusted parameters (filtering strength, spatial sigma, range sigma) based on local image characteristics such as gradient magnitude and variance. This allows the filter to adapt its behavior to different regions, strongly suppressing ringing artifacts in smooth areas while preserving edge structures in high-contrast regions, thereby achieving more complete artifact removal than fixed-parameter filters
Solution Approach 2:
The filtering process is made dynamic through conditional application based on gradient analysis. The patent calculates gradients in horizontal and vertical directions, applies thresholds to identify edge regions, and selectively applies the bilateral filter only in non-edge regions where ringing artifacts are present. This dynamic, region-dependent approach ensures complete artifact removal without compromising edge integrity
2Object-affected harmful factors
If a bilateral filter is applied to reduce ringing artifacts, then artifact removal improves, but computational complexity increases
Solution Approach 1:
The patent implements local quality analysis by computing gradient magnitudes and variances in different regions of the video block. The bilateral filter is applied with different parameters or not applied at all in different local regions - specifically, it is suppressed in high-gradient edge regions and applied in low-gradient smooth regions. This localized approach reduces overall computational complexity by avoiding unnecessary filtering operations in regions where they would not be beneficial
Solution Approach 2:
The video block is segmented into different regions based on gradient analysis - edge regions identified by high gradient magnitudes and smooth regions identified by low gradients. The bilateral filter is selectively applied to specific segments (smooth regions) while skipped in others (edge regions), thereby reducing the total number of filter operations and computational complexity while maintaining effective artifact removal in the regions where it is applied
3Object-affected harmful factors
If filtering parameters are adjusted to strongly remove artifacts, then artifact reduction improves, but edge structures may be blurred
Solution Approach 1:
The patent dynamically changes filter parameters based on local image characteristics. The filtering strength parameter is adjusted according to gradient magnitude - using strong filtering in smooth regions with low gradients and reducing or eliminating filtering in edge regions with high gradients. The spatial sigma and range sigma parameters are also adapted locally, allowing strong artifact suppression in homogeneous areas while preserving sharp transitions at edges
Solution Approach 2:
The patent employs feedback through gradient-based edge detection to control the filtering process. By calculating horizontal and vertical gradients and comparing them against thresholds, the system identifies edge locations and uses this information to modulate the bilateral filter application. This feedback mechanism ensures that filtering intensity is automatically reduced near edges, preventing edge blurring while maintaining strong artifact removal in non-edge regions
Data Source
AI summary
A mechanism for processing video data is disclosed. The mechanism determines to apply a bilateral filter to samples in a current block of a current picture. The bilateral filter includes filter weights that vary based on a distance between surrounding samples and a central sample and differences in intensities of the surrounding samples and the central sample. A conversion is performed between a visual media data and a bitstream based on the bilateral filter.