VVC Deblocking Filter Selection for Natural Texture Preservation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The current deblocking filter decisions in VVC do not adequately consider natural structures at sample points like p6 and q6, leading to potential degradation of image quality by removing natural textures.
Innovation Solution
Modify the long filter decision in VVC by adding gradient checks that include sample points p6 and q6, and adjust the dpq threshold from beta>>2 to beta>>4 to prevent unnecessary filtering of natural structures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If long filter is applied to block boundaries, then blocking artifacts are reduced, but natural structures and textures are removed or blurred
Solution Approach 1:
The patent applies different filtering strengths to different boundary locations by introducing gradient checks at specific sample points (p6, q6). The filter decision is localized based on whether natural structures are detected at these critical positions, rather than applying uniform filtering across the entire boundary. This allows strong filtering where blocking artifacts exist without affecting natural structures.
Solution Approach 2:
The patent performs gradient checks at sample points p6 and q6 before applying the long filter decision. By evaluating the gradient values at these positions in advance, the method determines whether natural structures are present and adjusts the filtering strength accordingly, preventing premature or inappropriate filtering that would degrade natural textures.
2Object-affected harmful factors
If gradient checks at p6 and q6 are added to long filter decision, then natural structures are preserved, but computational complexity increases
Solution Approach 1:
Instead of performing complex analysis across the entire boundary, the patent focuses gradient checks only on specific critical sample points (p6 and q6). This localized approach preserves natural structures while minimizing the increase in computational complexity by limiting the additional calculations to just these key positions.
3Manufacturing precision
If dpq threshold is reduced from beta>>2 to beta>>4, then filtering is applied more selectively, but more natural structures may be missed
Solution Approach 1:
The patent compensates for the lower threshold by introducing localized gradient checks at specific sample points p6 and q6. These position-specific checks ensure that natural structures are detected even with the reduced threshold, maintaining detection accuracy while allowing more selective filter application throughout the boundary.
Solution Approach 2:
The gradient values at sample points p6 and q6 act as intermediary indicators that bridge the gap between the reduced threshold and accurate natural structure detection. These intermediate measurements provide additional information to confirm whether filtering should be applied, preventing false positives that would occur with the lower threshold alone.
Data Source
AI summary
The long filter decision in VVC is modified by adding at least one gradient check that at least includes sample p6 or q6. This makes it possible to avoid using the long filters when there is some natural structure at sample p6 or q6. In one specific embodiment at least two gradient checks including both q6 and p6 are added. In another embodiment, the dpq threshold is modified from beta>>2 to beta>>4. This threshold change embodiment may be used in conjunction with or instead of the embodiment in which a gradient check that includes p6 and/or q6 is added to the long filter decision.