Virtual Reference Pictures for Video Coding Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video compression systems, such as the MPEG-4 AVC standard, rely on previously reconstructed/decoded pictures for prediction, which can be limited in enhancing prediction quality and coding efficiency due to global illumination mismatch and noise suppression.
Innovation Solution
The use of virtual reference pictures, which are processed and generated from decoded pictures, allowing for improved prediction signals and coding efficiency through methods like weighted prediction and adaptive reference generation, without requiring complete generated reference pictures for display purposes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If previously decoded pictures are used as reference for prediction, then the system maintains simplicity in reference management, but prediction quality is limited due to global illumination mismatch and noise
Solution Approach 1:
The patent applies preliminary action by generating virtual reference pictures in advance from already decoded pictures before they are needed for prediction. This preprocessing step creates enhanced reference pictures that address global illumination mismatch and noise issues, allowing the prediction process to use higher quality references without increasing real-time processing complexity
Solution Approach 2:
The patent introduces virtual reference pictures as an intermediary between the decoded pictures and the prediction process. These virtual reference pictures serve as a mediating element that improves prediction quality by addressing illumination mismatch and noise, while the system manages these intermediaries through signaling mechanisms that control their generation and usage
2Productivity
If weighted prediction and adaptive reference generation are applied to processed decoded pictures, then coding efficiency is improved, but the system complexity increases
Solution Approach 1:
The patent applies self-service by enabling the decoding system to generate its own enhanced reference pictures from the decoded output. The virtual reference picture generation process uses the decoded pictures themselves as input to create improved references, allowing the system to service its own prediction needs with enhanced quality without requiring external processing resources
Solution Approach 2:
The patent applies parameter changes by modifying the reference picture parameters through virtual reference picture generation. By changing the state of reference pictures from raw decoded output to processed virtual references with adjusted illumination and noise characteristics, the system improves prediction accuracy and coding efficiency while managing complexity through controlled parameter transformation
Data Source
AI summary
There are provided methods and apparatus using virtual reference pictures. An apparatus includes an encoder for encoding at least one picture, using at least one virtual reference picture, to form a resultant bitstream. The at least one virtual reference picture is different than a previously decoded picture.