HEVC Deblocking Filter Selection for Lower Complexity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing deblocking filter techniques in HEVC are too complex, leading to increased computational and memory requirements, which are exacerbated by the need to handle ultra-high-definition images, and do not effectively remove blocking artifacts.
Innovation Solution
A method is introduced to determine boundary strength for each 4-sample edge on an 8×8 sample grid, deciding whether deblocking filtering is necessary, selecting an appropriate filter, and applying it to reduce computational complexity and memory usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the existing deblocking filter technique in HEVC is used, then blocking artifacts can be removed, but the computational complexity and memory requirements increase significantly
Solution Approach 1:
The patent extracts and removes unnecessary boundary strength determination steps for certain block types (intra-block copy mode blocks, skip mode blocks, and blocks with specific prediction modes). By taking out these redundant operations, the computational complexity and memory requirements are reduced while maintaining the essential deblocking functionality for blocks that actually require it.
Solution Approach 2:
The patent applies deblocking filtering selectively based on local block characteristics. Instead of uniformly applying the complex boundary strength determination to all blocks, it tailors the filtering approach to specific block types and prediction modes, applying full complexity only where necessary and simplifying or skipping where not needed.
2Reliability
If the existing deblocking filter technique in HEVC is used, then blocking artifacts can be removed, but the processing time increases
Solution Approach 1:
The patent removes unnecessary boundary strength determination operations for specific block types including intra-block copy mode blocks, skip mode blocks, and blocks with certain prediction modes. This extraction of redundant processing steps directly reduces the total processing time while preserving essential deblocking functionality.
Solution Approach 2:
The patent applies deblocking filtering partially rather than universally. By determining boundary strength only for blocks that require it and skipping or simplifying the process for blocks where it is unnecessary, the processing time is reduced while maintaining adequate artifact removal for the blocks that need it.
3Reliability
If boundary strength is determined for all blocks in HEVC, then comprehensive deblocking is achieved, but memory capacity requirements increase
Solution Approach 1:
The patent extracts and eliminates the need to store boundary strength information for blocks that do not require deblocking filtering. By identifying and removing these unnecessary storage requirements for specific block types, memory capacity requirements are reduced while maintaining comprehensive deblocking for blocks that actually need it.
Solution Approach 2:
The patent applies boundary strength determination and associated memory storage selectively based on local block characteristics. Instead of uniformly storing boundary strength for all blocks, it tailors the storage requirements to specific block types and prediction modes, reducing overall memory capacity needs while maintaining deblocking coverage where necessary.
Data Source
AI summary
A method for processing image data can include determining a boundary strength per each 4-sample edge which is a boundary of a transform unit or a prediction unit and lies on an 8×8 sample grid, determining whether to apply a deblocking filtering on the 4-sample edge when the boundary strength of the 4-sample edge is not zero, selecting a deblocking filter to be applied on the 4-sample edge when it is determined to apply a deblocking filtering on the 4-sample edge, filtering the 4-sample edge by using the selected deblocking filter, and applying an edge offset when a sample adaptive offset (SAO) type indicates the edge offset.