MMVD Derivation for Image Decoding Bitstream Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing demand for high-resolution, high-quality images leads to increased data transmission and storage costs due to the high amount of information required, necessitating a more efficient image compression technique.
Innovation Solution
An image decoding method and apparatus that derive a merge motion vector difference (MMVD) based on MMVD index information and a predefined table, using integer or fractional sample precision, to reduce the bitstream size and improve coding efficiency, especially considering current picture referencing (CPR) for efficient image coding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-resolution, high-quality images are transmitted or stored using conventional methods, then image quality is maintained, but transmission cost and storage cost increase due to the large amount of data
Solution Approach 1:
The patent extracts only the essential motion information needed for prediction by using merge candidate lists and MMVD techniques. Instead of transmitting complete motion vectors for all blocks, only differential values (MMVD) are transmitted, extracting the minimum necessary data to maintain prediction accuracy while reducing overall bitstream volume.
Solution Approach 2:
The patent segments the motion information transmission by dividing it into merge candidate selection and MMVD refinement stages. The merge candidate list provides coarse motion information from neighboring blocks, while MMVD adds fine-grained adjustments. This segmentation allows progressive refinement of motion prediction with controlled bit allocation.
2Productivity
If conventional motion compensation methods are used, then motion prediction is performed, but the bitstream size remains large due to detailed motion vector transmission
Solution Approach 1:
The patent introduces merge candidates as intermediary elements that mediate between full motion vectors and simple copy modes. These candidates serve as intermediate representations that can be refined with MMVD, providing a middle ground that reduces bitstream size while maintaining prediction flexibility and accuracy.
Solution Approach 2:
The patent performs preliminary action by constructing merge candidate lists from neighboring blocks before final motion vector determination. This preliminary selection of candidate motion vectors from spatial and temporal neighbors allows the decoder to have a head start in motion compensation, reducing the information that needs to be transmitted in the bitstream.
Data Source
AI summary
A decoding apparatus performs an image decoding method by receiving a bit stream including prediction information of a current block; deriving motion information of the current block on the basis of a merge candidate indicated by a candidate flag of the current block in a merge candidate list; deriving a MVD of the current block on the basis of MVD information of the current block and whether or not integer sample precision or fractional sample precision is used in the motion information; deriving modified motion information of the current block on the basis of the motion information and the MVD; and performing prediction of the current block on the basis of the modified motion information.