Image Decoding MMVD Precision Control to Reduce Syntax Overhead
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In Versatile Video Coding (Draft 9), JVET-R2001, the ph_mmvd_fullpel_only_flag is unnecessarily decoded even when the transmission precision of MMVD is specified to be integer precision, leading to unnecessary syntax decoding processing and increased bit volumes.
Innovation Solution
An image decoding device and method that decodes first syntax to control MMVD transmission precision in a sequence unit, using integer or fractional precision as needed, reducing unnecessary decoding and bit volumes by only decoding syntax when the precision is not already determined.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If ph_mmvd_fullpel Only flag is decoded in all cases, then MMVD transmission precision can be controlled in picture unit, but unnecessary syntax decoding processing and increased bit volumes occur when sequence unit already specifies integer precision
Solution Approach 1:
The decoder first checks the sps_mmvd_to_flag in the sequence parameter set to determine whether integer precision is specified at the sequence level. Based on this preliminary information, the decoder decides whether to further decode the ph_mmvd_to_flag in the picture header. This prevents unnecessary syntax decoding when the sequence level already specifies integer precision, reducing processing complexity and bit volume while maintaining the ability to control MMVD transmission precision when needed.
2Measurement precision
If ph_mmvd_to_flag is always decoded, then picture-level precision control is available, but decoding processing volume and bit volumes increase unnecessarily
Solution Approach 1:
The decoder performs a preliminary check of the sps_mmvd_to_flag before decoding the ph_mmvd_to_flag. When the sequence parameter set indicates integer precision is already specified, the picture-level flag is skipped, avoiding unnecessary decoding operations and reducing processing time while preserving the option to decode picture-level precision control when the sequence level does not specify precision.
3Adaptability or versatility
If ph_mmvd_to_flag is decoded in all pictures, then precise control of MMVD transmission precision per picture is achieved, but bit volumes increase unnecessarily
Solution Approach 1:
The decoder checks the sps_mmvd_to_flag in the sequence parameter set as a preliminary step. When this flag indicates integer precision is specified at the sequence level, the decoder skips decoding the ph_mmvd_to_flag in the picture header, thereby reducing the bit volume required for picture-level precision control while maintaining the capability to specify picture-level precision when the sequence level does not already determine it.
Data Source
AI summary
An image decoding device according to the present invention including a decoding unit configured to decode first syntax that controls transmission precision of a merge with motion vector difference in a sequence of a decoding target, wherein the value of the first syntax being “1” specifies that both fractional precision and integer precision could be used as the transmission precision of the merge with motion vector difference in the sequence of the decoding target, and the value of the first syntax being “0” specifies that the first syntax uses fractional precision as the transmission precision of a merge with motion vector difference in the sequence of the decoding target.