Parallel Inter-View Motion Vector Prediction for Multiview Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video signal processing techniques face inefficiencies in coding multiview video images, particularly in deriving accurate motion vectors for prediction, which affects the restoration and compression of video data.
Innovation Solution
The method involves acquiring an inter-view motion vector of a current coding unit through parallel processing and using this vector to obtain motion vectors for prediction units, leveraging picture order counts and motion information from corresponding blocks across different views to enhance prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional sequential processing is used to derive motion vectors for multiview video coding, then processing accuracy can be maintained, but processing speed and coding efficiency deteriorate
Solution Approach 1:
The patent divides the current picture into multiple parallel processing units, each handling a specific spatial region. Within each unit, prediction units are processed independently using segmented motion vector derivation methods, allowing simultaneous processing of multiple regions while maintaining accurate motion compensation through view-specific reference picture selection.
Solution Approach 2:
The patent introduces a new dimension of parallel processing by simultaneously processing multiple prediction units within a coding unit. This is achieved by deriving motion vectors for different prediction units in parallel using inter-view motion vector prediction, where motion information from corresponding blocks in other views is utilized to enable independent computation without sequential dependencies.
2Productivity
If inter-view motion vector prediction is implemented for all prediction units, then coding efficiency improves, but processing complexity increases
Solution Approach 1:
The patent applies inter-view motion vector prediction selectively based on local characteristics of each prediction unit. The method determines whether to use inter-view prediction for each prediction unit individually by checking if a corresponding block exists in another view and if the motion vector derivation is beneficial, thereby optimizing coding efficiency while avoiding unnecessary processing complexity in regions where inter-view prediction is not applicable.
3Loss of time
If parallel processing is used to speed up video restoration, then processing time decreases, but motion vector prediction accuracy may deteriorate
Solution Approach 1:
The patent performs preliminary derivation of inter-view motion vectors from corresponding blocks in other views before processing the current prediction unit. By pre-computing these reference motion vectors and storing them for use in parallel processing, the method enables simultaneous processing of multiple prediction units while maintaining accurate motion compensation, as the reference information is already prepared and available.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
According to the present invention, an inter-view motion vector of a current coding unit is acquired using parallel processing information, and a motion vector of a current prediction unit in the current coding unit is acquired in parallel using the inter-view motion vector of the current coding unit. The present invention relates to a method and device for processing a video signal, wherein a motion vector prediction value of the current prediction unit is acquired by comparing output order information of a reference picture corresponding to the current prediction unit and output order information of a reference picture corresponding to a corresponding block. According to the present invention, through a parallel processing step, a multi-view point video image can be quickly restored, and the accuracy of the motion vector prediction value can be increased using motion information corresponding to the reference picture of the corresponding block of a different view point from the current prediction unit.