Adaptive Motion Vector Prediction for Video Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In 'Versatile Video Coding (Draft 7)', the requirement that the index value for the collocated picture is the same for all slices in a coded picture leads to redundancy when multiple slices are present, necessitating redundant designation of the same value multiple times.
Innovation Solution
Adaptive motion vector prediction is implemented, allowing the collocated picture to be designated on a per picture basis and the reference list on a per slice basis, reducing redundancy by enabling different index values for each slice.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the index value for the collocated picture is designated on a per slice basis as required by the standard, then the collocated picture can be correctly identified for temporal prediction, but the same value is designated multiple times across slices leading to redundancy
Solution Approach 1:
The patent segments the designation of collocated picture index into two parts: picture-level designation (common to all slices) and slice-level reference list designation. The picture-level index designates the collocated picture once for the entire picture, while slice-level parameters only designate reference lists, avoiding redundant repetition of the same index value across multiple slices.
Solution Approach 2:
The patent merges the picture-level collocated picture index designation with slice-level reference list designation into a unified system. The picture-level index provides the base collocated picture identification, while slice-level parameters provide reference list selections, combining both levels to achieve complete functionality without redundancy.
2Adaptability or versatility
If the collocated picture index is designated for each slice individually, then each slice can have optimal prediction parameters, but the overall coding efficiency decreases due to redundant data transmission
Solution Approach 1:
The patent applies local quality by allowing slice-specific reference list designation while maintaining picture-level collocated picture consistency. Each slice can select from reference lists that are locally optimized for its specific characteristics, while the underlying collocated picture index remains uniform across all slices, achieving both adaptability and efficiency.
3Reliability
If redundant designation of the same collocated picture index is performed across multiple slices, then the standard requirement is satisfied, but the data size and processing load increase
Solution Approach 1:
The patent extracts the collocated picture index designation from the slice-level parameters and places it at the picture level. This extraction eliminates the need to repeat the same index value in each slice's header, reducing data size while maintaining standard compliance through the preserved picture-level designation.
Data Source
AI summary
An adaptive motion vector prediction unit configured to adaptively perform spatial prediction that performs prediction using a motion vector around a target block and temporal prediction that performs prediction using a motion vector of a collocated picture is included, and in the temporal prediction performed by the adaptive motion vector prediction unit, the collocated picture to be referred to is designated on a per picture basis, and a reference list is designated on a per slice basis.