SAO Filtering for 360-Degree VR Projection Frames
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing omnidirectional video content processing techniques face challenges in accurately performing sample adaptive offset (SAO) filtering, particularly at image content discontinuities in 360-degree virtual reality (VR) projection-based frames, leading to degraded picture quality due to lack of neighboring pixels and inefficient handling of face and picture boundaries.
Innovation Solution
A padding-based SAO filtering method is employed, where padding pixels are used to extend face and picture boundaries, enabling accurate SAO filtering by classifying and calculating offsets for pixels near discontinuities, thereby improving image quality by utilizing padding pixels as neighboring references.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional SAO filtering is applied to projection-based frames with image content discontinuity, then processing speed is maintained, but filtering accuracy deteriorates at face boundaries and picture boundaries due to lack of valid neighboring pixels
Solution Approach 1:
The patent segments the projection-based frame into multiple projection faces and identifies boundary regions between them. By separating boundary pixels from interior pixels, the method applies different processing strategies: interior pixels use conventional SAO filtering while boundary pixels receive special handling through padding pixel generation, thus resolving the accuracy issue at discontinuities without overwhelming complexity
Solution Approach 2:
The patent introduces padding pixels as intermediary elements at face boundaries and picture boundaries. These padding pixels act as mediators that provide the necessary neighboring pixel values for SAO filtering at boundary locations where valid neighbors are otherwise unavailable, thereby maintaining filtering accuracy without requiring complex boundary detection and handling logic
2Measurement precision
If padding pixels are generated for all boundary pixels, then filtering accuracy at boundaries is improved, but computational complexity and processing time increase
Solution Approach 1:
The patent applies padding pixel generation selectively only to boundary pixels located at face boundaries and picture boundaries, rather than uniformly to all pixels in the frame. This localized approach ensures that filtering accuracy is improved precisely where needed (at discontinuities) while minimizing the additional computational burden, as the majority of interior pixels continue to use the efficient conventional SAO filtering process
Solution Approach 2:
The patent generates padding pixels with sufficient redundancy to ensure all boundary filtering needs are met, applying the padding strategy more extensively than the minimum required. This partial excessive action ensures that even complex boundary cases are handled correctly, and the resulting improvement in boundary filtering accuracy outweighs the moderate increase in processing overhead
Data Source
AI summary
A sample adaptive offset (SAO) filtering method for a reconstructed projection-based frame includes: obtaining at least one padding pixel in a padding area that acts as an extension of a face boundary of a first projection face, and applying SAO filtering to a block that has at least one pixel included in the first projection face. In the reconstructed projection-based frame, there is image content discontinuity between the face boundary of the first projection face and a face boundary of a second projection face. The at least one padding pixel is involved in the SAO filtering of the block.