Inter-layer Reference Picture Construction for Spatial Scalability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In scalable video coding (SVC) within the High Efficiency Video Coding (HEVC) framework, the challenge arises when the resolution of an upsampled reference layer picture does not match the resolution of the enhancement layer picture, hindering direct use as an inter-layer reference picture, and signaling of upsampled regions is complex, especially when the enhancement layer is predicted from multiple layers.
Innovation Solution
The method involves upsampling at least a portion of the reference layer picture to an upsampled picture with specified horizontal and vertical sizes, using position information relative to the enhancement layer to generate an inter-layer reference picture, which can be adjusted by padding or cropping to match the enhancement layer's resolution, and signaling this information to align the upsampled region with the enhancement layer's coordinates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the reference layer picture is upsampled to match the enhancement layer resolution, then inter-layer prediction accuracy is improved, but the coding complexity increases due to resolution mismatch and signaling requirements
Solution Approach 1:
The reference layer picture is divided into multiple tiles, and only the tiles that correspond to the enhancement layer picture region are upsampled. This segmentation approach reduces the amount of data that needs to be processed and transmitted, thereby lowering coding complexity while maintaining prediction accuracy for the relevant regions.
Solution Approach 2:
Instead of uniformly upscaling the entire reference layer picture, the patent applies upsampling only to specific local regions (tiles) that are needed for the enhancement layer prediction. This local quality approach ensures high prediction accuracy where needed while avoiding unnecessary processing elsewhere, thus reducing overall coding complexity.
2Ease of operation
If the upsampled reference layer picture is used directly, then inter-layer prediction is simplified, but resolution mismatch prevents direct use as reference picture
Solution Approach 1:
The enhancement layer picture is divided into tiles that correspond to specific regions in the upsampled reference layer picture. By processing and transmitting only the relevant tile data rather than the entire picture, the patent achieves resolution matching for the needed regions while simplifying the overall inter-layer prediction process.
Solution Approach 2:
The patent introduces a tile-based dimensional organization where the reference layer and enhancement layer are matched not through full-picture resolution scaling but through corresponding tile regions. This dimensional change from full-picture to tile-based matching simplifies the resolution alignment process while maintaining precision where needed.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method of coding video data includes upsampling at least a portion of a reference layer picture to an upsampled picture having an upsampled picture size. The upsampled picture size has a horizontal upsampled picture size and a vertical upsampled picture size. At least one of the horizontal or vertical upsampled picture sizes may be different than a horizontal picture size or vertical picture size, respectively, of an enhancement layer picture. In addition, position information associated with the upsampled picture may be signaled. An inter-layer reference picture may be generated based on the upsampled picture and the position information.