Video Coding With Diagonal Motion Merge Candidates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding techniques face challenges in efficiently compressing high-resolution images and 3D video content, particularly in terms of improving encoding/decoding efficiency and managing motion prediction and compensation.
Innovation Solution
The method involves dividing a current block into two partitions through diagonal motion division, constructing a merge candidate list, and using weighted prediction based on encoding parameters to enhance inter prediction, thereby improving encoding/decoding efficiency and reducing memory bandwidth.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If diagonal motion division is applied to divide current block into two partitions, then inter prediction accuracy is improved, but device complexity increases due to additional partitioning logic and merge candidate list construction
Solution Approach 1:
The current block is divided into two partitions through diagonal motion division, where each partition can independently select motion vectors from the merge candidate list. This segmentation allows different regions to use different prediction directions, improving prediction accuracy for blocks with complex motion patterns while maintaining manageable complexity through standardized partitioning logic.
Solution Approach 2:
The merge candidate list is adaptively constructed based on encoding parameters such as slice type, inter mode, block size/shape, and width-height ratio. This dynamic adaptation allows the system to optimize the merge candidate selection process for different block characteristics, improving prediction accuracy without requiring complex fixed logic for all cases.
2Measurement precision
If merge candidate list is constructed with multiple candidates including spatial, temporal, and combined candidates, then motion information accuracy is improved, but loss of time increases due to extended processing for list construction
Solution Approach 1:
The merge candidate list is constructed in advance during the encoding process, organizing spatial merge candidates, temporal merge candidates, and combined merge candidates into a structured list. This preliminary organization allows the decoder to efficiently select motion information without performing complex searches during decoding, reducing decoding time while maintaining high motion information accuracy.
Solution Approach 2:
Motion information from neighboring blocks (spatial candidates) and reference frames (temporal candidates) is copied and stored in the merge candidate list. This copying mechanism allows rapid retrieval of proven motion vectors without re-computation, significantly reducing processing time while maintaining accuracy through reuse of validated motion data.
3Measurement precision
If weighted prediction is applied to pixels near division boundaries, then prediction accuracy is improved, but use of energy increases due to additional computational operations
Solution Approach 1:
Weighted prediction is applied selectively to pixels near the division boundaries rather than uniformly across the entire block. The weighting scheme uses coefficients such as [7:1], [6:2], [4:4], [2:6], or [1:7] depending on the pixel's distance from the boundary. This local application improves prediction accuracy at critical boundary regions while minimizing additional computational energy compared to full-block weighted prediction.
Data Source
AI summary
In a video encoding/decoding method and device according to the present invention, a merge candidate of a current block may be determined, a merge candidate list may be configured on the basis of the merge candidate, motion information of the current block may be derived from the merge candidate list, and the motion information may be used to perform inter prediction of the current block.