Virtual Boundary Signaling for Loop Filtering Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In video encoding, discontinuous boundaries between faces in a frame-packed picture can lead to visible face seam artifacts when in-loop filtering is performed across these discontinuities.
Innovation Solution
The method involves receiving video data containing faces, selecting virtual boundaries between these faces, and disabling in-loop filtering across the selected virtual boundaries to prevent seam artifacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If in-loop filtering is performed across face boundaries, then compression efficiency is improved, but visible seam artifacts are generated at discontinuous boundaries
Solution Approach 1:
The patent applies different filtering treatments to different regions: in-loop filtering is applied within face regions to improve compression efficiency, while filtering is disabled or modified at discontinuous boundaries between faces to prevent seam artifacts. This local differentiation resolves the contradiction by optimizing each region according to its specific characteristics.
Solution Approach 2:
The patent segments the video picture into multiple face regions and identifies discontinuous boundaries between them. By segmenting the filtering operation into regions where filtering is applied and regions where it is disabled (at boundaries), the patent simultaneously achieves compression efficiency within faces while avoiding artifacts at boundaries.
2Manufacturing precision
If in-loop filtering is applied to all regions, then overall video quality is improved, but artifacts appear at discontinuous boundaries between faces
Solution Approach 1:
The patent implements spatially adaptive filtering where the filtering operation is selectively applied based on location. Within continuous face regions, filtering improves video quality by reducing noise and artifacts. At discontinuous boundaries between faces, filtering is disabled to prevent the creation of unwanted boundary artifacts, thus achieving overall quality improvement without introducing harmful effects.
Solution Approach 2:
The patent introduces dynamic control of the in-loop filtering process by detecting face boundaries and adjusting filtering application in real-time during video encoding. The filtering is dynamically enabled within faces and disabled at boundaries, allowing the system to adapt to local image characteristics and avoid artifacts while maintaining quality.
3Object-affected harmful factors
If in-loop filtering is disabled at all boundaries, then seam artifacts are prevented, but compression efficiency and video quality deteriorate
Solution Approach 1:
Instead of uniformly disabling filtering across the entire picture, the patent applies filtering locally within face regions while disabling it only at discontinuous boundaries. This selective approach prevents seam artifacts at boundaries while preserving the compression efficiency and quality benefits of filtering within the face regions.
Solution Approach 2:
The patent applies in-loop filtering partially - not across the entire picture but only in regions where it is beneficial (within faces). This partial application is sufficient to prevent seam artifacts while maintaining adequate compression efficiency and video quality within the face regions, avoiding the need to apply filtering excessively across all boundaries.
Data Source
AI summary
A method, computer program, and/or computer system is provided for coding a video sequence. Video data containing one or more faces may be received. One or more virtual boundaries between the one or more faces of the received video data may be selected. In-loop filtering may be disabled between the one or more selected virtual boundaries from among the virtual boundaries, based on the selected boundaries being discontinuous in a three-dimensional geometry.