Temporal Motion Vector Prediction in Scalable Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Scalable video coding systems, such as HEVC, face inefficiencies in processes like temporal motion vector predictors and reference list selection, which affect coding performance.
Innovation Solution
The method involves deriving temporal motion vector prediction candidates and selecting reference lists for the enhancement layer by considering motion vector scaling operations, ensuring that motion vector scaling is avoided when possible, and prioritizing temporal candidates over spatial ones in the candidate list construction for improved prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If motion vector scaling operation is performed in scalable video coding, then motion vector prediction can be adapted to different temporal resolutions, but coding efficiency and prediction accuracy deteriorate due to loss of precision
Solution Approach 1:
The patent extracts and removes the motion vector scaling operation from the prediction process by selecting reference pictures with matching temporal characteristics (same POC distance). This eliminates the need for scaling while maintaining adaptability to different temporal resolutions, thereby preserving motion vector prediction accuracy without sacrificing versatility.
2Measurement precision
If temporal candidates are prioritized in candidate list construction, then prediction accuracy improves for enhancement layer, but device complexity increases due to additional reference list selection processes
Solution Approach 1:
The patent performs preliminary organization of reference picture lists during the encoding process, pre-establishing lists grouped by temporal characteristics (POC distances). This preliminary action enables the decoder to efficiently select appropriate reference pictures without complex real-time calculations, thereby improving prediction accuracy while minimizing additional device complexity.
3Measurement precision
If multiple candidate reference picture lists are examined to avoid motion vector scaling, then motion vector prediction accuracy improves, but coding time and processing overhead increase
Solution Approach 1:
The patent segments the reference picture list into multiple sub-lists organized by temporal characteristics (different POC distances). This segmentation allows the encoder to quickly identify and select the appropriate segment without examining all candidate reference pictures, thereby improving prediction accuracy while reducing coding time and processing overhead through targeted search.
Data Source
AI summary
In enhancement layer (EL) coding, the temporal candidate(s) may be derived the merge mode and the non-merge mode. Selection of the reference index of a current prediction unit (PU) and of the reference list of a co-located PU may be considered jointly when constructing the temporal candidate of EL for the merge mode. The selection^) may be made such that a motion vector (MV) scaling operation may be avoided. A selected motion vector prediction candidate may be added to a merging candidate list in a position before that of a spatial motion vector prediction candidate. A selected motion vector prediction candidate may be added to the merging candidate list in place of a spatial motion vector prediction candidate.