Inter-view Candidate Derivation for 3D Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current multi-view video coding techniques face inefficiencies due to high storage and transmission requirements, as straightforward approaches fail to effectively exploit inter-view redundancy, leading to poor coding performance.
Innovation Solution
Incorporating a second inter-view candidate into the candidate list for Merge, Skip, and AMVP-based Inter modes, derived from motion information associated with a corresponding block in a reference view, to enhance coding efficiency by improving the selection of motion and disparity vectors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional video coding techniques are applied to each single-view video sequence independently, then device complexity is reduced, but coding performance deteriorates due to failure to exploit inter-view redundancy
Solution Approach 1:
The patent merges the candidate list construction processes of both views by allowing each view to use motion vectors from corresponding blocks in other views as candidates. This combining of candidate sources exploits inter-view redundancy while maintaining a unified coding framework, thereby improving coding performance without proportionally increasing device complexity
Solution Approach 2:
The patent creates a universal candidate list construction mechanism that works for both intra-view and inter-view prediction. The same candidate list structure and selection process is used across different views, making the system multi-functional and adaptable to various prediction scenarios without requiring separate complex processing paths
2Reliability
If the candidate list size is increased to improve motion vector prediction accuracy, then coding efficiency improves, but processing complexity increases
Solution Approach 1:
The patent applies partial action by selectively adding inter-view candidates to the candidate list based on specific conditions (e.g., when inter-view redundancy is beneficial). Rather than always maximizing candidate list size, the system adds candidates partially and conditionally, improving coding efficiency when needed while avoiding unnecessary processing complexity when inter-view candidates would not provide benefit
Solution Approach 2:
The patent dynamically changes the candidate list parameters (size, composition) based on the specific coding context. The candidate list is adapted by incorporating inter-view candidates when they provide predictive value, and the system adjusts the candidate selection parameters to balance between list size and processing complexity, thereby optimizing coding efficiency without excessive complexity increase
Data Source
AI summary
A method and apparatus of three-dimensional/multi-view coding using a candidate list including a second inter-view candidate in the candidate list for Merge mode, Skip mode or AMVP based (Advanced Motion Vector Prediction based) Inter mode are disclosed. The second inter-view candidate can be derived based on already coded or decoded texture data for the candidate list to include. For example, the second inter-view candidate can be determined from the motion information associated with a corresponding block in a reference view, where the corresponding block is located according to the location of the right-bottom neighboring block and a selected disparity vector. The right-bottom neighboring block is located across from a right-bottom corner of the current texture block. The second inter-view candidate can be inserted into the candidate list only when the number of previous available candidates is smaller than a pre-specified number.