Gradual Decoding Refresh for Clean-Region Intra Mode Propagation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video compression systems face inefficiencies in handling video encoding and decoding processes, particularly in managing intra mode propagation and motion vector handling across clean and dirty regions in reference pictures.
Innovation Solution
The proposed solution involves a video encoder that distinguishes between clean and dirty regions in reference pictures, using intra mode propagation information and motion compensated blocks to determine validity and discard candidate modes or vectors based on their location relative to boundaries, thereby optimizing encoding decisions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If intra mode propagation is performed across the entire reference picture, then encoding efficiency is improved, but reliability deteriorates due to propagation of errors from dirty regions
Solution Approach 1:
The reference picture is segmented into clean and dirty regions by a GDR boundary. The encoder checks the location of intra-coded blocks relative to this boundary and only propagates intra mode information from clean regions, preventing error propagation from dirty regions while maintaining encoding efficiency in valid areas.
Solution Approach 2:
Different regions of the reference picture are treated differently based on their quality. Clean regions are allowed to propagate intra mode information to improve encoding efficiency, while dirty regions have their propagation restricted to maintain reliability. The intra mode propagation validity indication is set locally based on the specific block's location relative to the GDR boundary.
2Measurement precision
If all candidate motion vectors are considered for encoding, then encoding accuracy is improved, but processing complexity increases due to validation requirements
Solution Approach 1:
The encoder performs preliminary checks on candidate motion vectors by determining the location of associated intra-coded blocks relative to the GDR boundary before full encoding processing. This preliminary validation filters out invalid candidates from dirty regions, reducing processing complexity while maintaining encoding accuracy for valid candidates.
Solution Approach 2:
The validity of candidate motion vectors is determined by changing the evaluation parameter based on location. Candidates associated with blocks in clean regions are marked as valid, while those in dirty regions are marked as invalid. This parameter-based filtering simplifies the complexity-accuracy tradeoff by using a clear location-based criterion.
3Productivity
If motion compensated blocks from dirty regions are used, then compression ratio is improved, but quality deteriorates due to unreliable reference data
Solution Approach 1:
The reference picture is divided into clean and dirty regions by a GDR boundary. Motion compensated blocks are evaluated based on their location: blocks from clean regions are used to improve compression ratio, while blocks from dirty regions are discarded to maintain quality, preventing the use of unreliable reference data.
Solution Approach 2:
Different quality standards are applied locally based on region. In clean regions, motion compensated blocks are accepted for encoding to achieve better compression. In dirty regions, blocks are rejected regardless of compression potential, ensuring that quality is maintained by only using reliable reference data from clean areas.
Data Source
AI summary
Systems, methods, and instrumentalities are disclosed herein for the field of video compression. Examples herein aim at improving compression efficiency compared to existing video compression systems. In examples, a video encoder may obtain a reference picture of a current picture. The reference picture may include a clean region and a dirty region divided by a boundary. The encoder may obtain a motion compensated block associated with a current block in the reference picture. The encoder may obtain, based on the motion compensated block, an associated intra-coded block. The encoder may determine intra mode propagation information for the current block based on a location of the intra-coded block relative to the boundary. The current picture may be encoded based on the intra mode propagation information.