Inter-layer Prediction in Scalable Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Scalable video coding technologies face challenges in efficiently utilizing base layer (BL) reconstructed video data for inter-layer prediction, leading to suboptimal coding efficiency and image quality, especially in adaptive bit-rate and resolution scenarios.
Innovation Solution
The method involves using the BL reconstructed texture frame or up-sampled BL reconstructed texture frame as an inter-layer reference picture for inter prediction, limiting motion vectors to a range around (0,0), and applying weighted prediction and adaptive filtering to improve inter-layer prediction accuracy and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If BL reconstructed video data is used for inter-layer prediction, then coding efficiency is improved, but prediction accuracy deteriorates due to suboptimal utilization
Solution Approach 1:
The patent applies parameter changes by introducing weighted prediction coefficients and adaptive filtering parameters to transform the base layer reconstructed video data into an enhanced prediction signal. The weighting mechanism adjusts the contribution of different prediction sources dynamically, while adaptive filtering modifies the statistical characteristics of the prediction signal to better match the enhancement layer content, thereby resolving the contradiction between coding efficiency and prediction accuracy.
Solution Approach 2:
The patent introduces an intermediary processing stage that sits between the base layer reconstructed video data and the final inter-layer prediction. This intermediary includes up-sampling operations, adaptive filtering, and weighted combination mechanisms that transform the base layer data into an optimized prediction signal, acting as a mediator that improves prediction accuracy while maintaining coding efficiency.
2Measurement precision
If adaptive filtering and weighted prediction are applied, then prediction accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent applies partial action by selectively applying adaptive filtering and weighted prediction only to specific blocks or regions where they provide the most benefit, rather than uniformly across the entire video frame. This selective application maintains prediction accuracy improvement while reducing overall computational complexity by avoiding unnecessary processing in areas where simple prediction suffices.
3Productivity
If motion vectors are limited to range around (0,0), then inter-layer prediction efficiency is improved, but adaptability to large motions deteriorates
Solution Approach 1:
The patent resolves this contradiction by introducing another dimension to the prediction process through weighted prediction and adaptive filtering. Instead of relying solely on motion vector magnitude, the invention adds a weighting dimension that allows accurate prediction even when motion vectors are constrained to small ranges, thereby maintaining both efficiency and adaptability.
Data Source
Figure 1
Figure 2~4
Figure 5
AI summary
A method and apparatus for inter-layer prediction for scalable video coding are disclosed. Embodiments of the present invention utilize weighted prediction for scalable coding. The weighted prediction is based on the predicted texture data and the inter-layer Intra prediction data derived from BL reconstructed data. The inter-layer Intra prediction data corresponds to the BL reconstructed data or up-sampled BL reconstructed data. The predicted texture data corresponds to spatial Intra prediction data or motion-compensated prediction data based on the second EL video data in the same layer as the current EL picture. Embodiments of the present invention also utilize the reference picture list including an inter-layer reference picture (ILRP) corresponding to BL reconstructed texture frame or up-sampled BL reconstructed texture frame for Inter prediction of EL video data. The motion vector is limited to a range around (0,0) when the ILRP is selected as a reference picture.