Adaptive Loop Filtering With Boundary Padding for Video Units
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding standards face challenges in handling virtual boundaries of coding tree blocks, leading to inefficiencies in compression and decoding processes, particularly in the application of adaptive loop filters across these boundaries.
Innovation Solution
The proposed methods involve determining the availability of neighboring samples across video unit boundaries and using padding techniques to replace unavailable samples, applying adaptive loop filtering operations based on predefined rules that ensure samples do not cross these boundaries, and utilizing specific filter coefficients and clipping parameters to maintain efficient conversion between video blocks and bitstream representations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If adaptive loop filtering is applied across video unit boundaries using neighboring samples, then filtering quality is improved, but complexity of handling boundary conditions increases
Solution Approach 1:
The video picture is divided into separate video units with defined boundaries. The filtering process is segmented to operate independently within each video unit, avoiding cross-boundary operations. This segmentation resolves the contradiction by eliminating the need to handle complex boundary conditions while maintaining filtering quality through unit-independent processing.
Solution Approach 2:
Padding samples are introduced as intermediary elements to replace unavailable neighboring samples at video unit boundaries. These padding samples act as mediators that allow the filtering process to proceed without directly accessing samples from adjacent video units, thus simplifying boundary handling while maintaining filtering continuity.
2Duration of action of stationary object
If padding samples are used to replace unavailable neighboring samples, then filtering operation continuity is improved, but accuracy of boundary sample representation deteriorates
Solution Approach 1:
The availability status of samples is changed as a parameter to control filtering behavior. By determining whether neighboring samples are available or unavailable based on video unit boundaries, the system adjusts the filtering process dynamically. This parameter change allows continuous filtering operations while managing the trade-off in boundary sample accuracy through controlled use of padding samples.
3Productivity
If video pictures are divided into multiple video units, then processing efficiency is improved, but number of virtual boundaries increases
Solution Approach 1:
The video picture is segmented into multiple independent video units, which improves processing efficiency by enabling parallel and independent processing of each unit. Although this creates more boundaries, the segmentation principle resolves the contradiction by making each boundary a simple, well-defined separation that requires minimal handling complexity, thus maintaining high processing efficiency.
Data Source
AI summary
A method of video processing includes determining, for a conversion between a current block of a video and a bitstream representation of the video using a coding tool that accesses samples outside of the current block, that neighboring samples used for the conversion are unavailable. Samples that are located across a boundary of a video unit of the video are considered as unavailable samples. The method also includes performing, based on the determining, the conversion by padding samples in place of the neighboring samples that are unavailable. The padding samples are determined using samples that are restricted to be within a current processing unit associated with the current block.