Disparity Vector Prediction in Multiview Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding techniques face challenges in efficiently determining disparity vectors for inter-view motion prediction in multiview and 3D video coding, particularly in deriving depth maps and relying on global disparity vectors, which are processing-intensive and do not accurately represent the disparity between corresponding blocks in different views.
Innovation Solution
The techniques described determine disparity vectors based on motion information of spatial and temporal neighboring blocks on a region-by-region basis, without the need to derive depth maps or rely on global disparity vectors, allowing for efficient inter-view motion prediction and accurate representation of block disparities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If depth maps are derived and global disparity vectors are used for inter-view motion prediction, then disparity vectors can be obtained for multiview and 3D video coding, but the processing becomes intensive and accuracy is reduced
Solution Approach 1:
The patent divides the video picture into multiple regions and determines disparity vectors for each region separately based on motion information from spatial and temporal neighboring blocks. This segmentation approach eliminates the need for global depth map derivation while maintaining accurate disparity representation for each local region, thereby resolving the contradiction between accuracy and processing complexity.
Solution Approach 2:
The patent applies local quality by determining disparity vectors on a region-by-region basis rather than using a single global disparity vector for the entire picture. Each region's disparity vector is derived from its specific motion information, providing locally accurate disparity representation without the intensive processing required for global depth map derivation.
2Productivity
If region-by-region disparity vector determination is performed without depth map derivation, then processing requirements are reduced, but the accuracy of disparity representation may be compromised
Solution Approach 1:
The patent performs preliminary action by utilizing motion information from spatial and temporal neighboring blocks that has already been computed during standard video coding. This pre-existing motion information is repurposed to determine disparity vectors, eliminating the need for separate depth map derivation while maintaining accuracy through the use of reliable neighboring block data.
Solution Approach 2:
The patent uses copying by taking motion vectors from spatial and temporal neighboring blocks and adapting them as disparity vectors for the current region. This copying approach leverages the already-computed motion information from neighboring blocks, providing accurate disparity representation without requiring additional intensive processing for depth map generation.
Data Source
AI summary
Techniques are described for determining a disparity vector for a current block based on disparity motion vectors of one or more spatially and temporally neighboring regions to a current block to be predicted. The spatially and temporally neighboring regions include one or a plurality of blocks, and the disparity motion vector represents a single vector in one reference picture list for the plurality of blocks within the spatially or temporally neighboring region. The determined disparity vector could be used to coding tools which utilize the information between different views such as merge mode, advanced motion vector prediction (AMVP) mode, inter-view motion prediction, and inter-view residual prediction.