Motion Vector Prediction List Construction for Video Coding Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video coding techniques have low coding efficiency due to ineffective motion vector prediction (MVP) list construction methods that do not fully exploit available information during encoding/decoding, leading to mismatches between true motion information and constructed MVP lists, limited diversity in MVP lists, and neglecting valuable motion information from non-adjacent temporal positions.
Innovation Solution
The proposed solution involves constructing an optimized MVP list using template matching costs to prioritize MVP candidates, incorporating non-adjacent temporal MVP candidates, and applying an adaptive reordering merge candidates (ARMC) process to improve the selection and ordering of MVP candidates, thereby enhancing the coding effectiveness and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional MVP list construction methods are used, then the encoding process is simple, but the coding efficiency is low
Solution Approach 1:
The patent performs preliminary actions by constructing multiple groups of MVP candidates and calculating template matching costs before the actual encoding process. This preliminary preparation allows the system to sort and prioritize candidates based on their matching costs, ensuring that the most relevant motion vector predictions are available when needed, thereby improving coding efficiency without excessive complexity during the main encoding process.
Solution Approach 2:
The patent segments the MVP candidate pool into multiple groups based on different sources (spatial neighbors, temporal neighbors, non-adjacent temporal positions). Each group is constructed and evaluated separately using template matching costs, allowing the system to systematically organize and prioritize candidates from different spatial and temporal contexts, improving both diversity and efficiency.
2Reliability
If MVP candidates are not sorted by template matching costs, then the construction process is faster, but the appropriateness of the MVP list is reduced
Solution Approach 1:
The patent changes the ordering parameter of MVP candidates from arbitrary or simple spatial/temporal proximity to template matching costs. By using template matching costs as the sorting criterion, the system ensures that candidates are arranged in descending order of their relevance to the current block, improving the appropriateness and reliability of the MVP list while managing construction time through efficient cost calculation methods.
3Adaptability or versatility
If only adjacent temporal MVP candidates are used, then the list construction is simpler, but the diversity of motion information is limited
Solution Approach 1:
The patent extends the temporal dimension by incorporating non-adjacent temporal MVP candidates in addition to adjacent temporal neighbors. This dimensional expansion allows the system to capture motion information from multiple temporal references, including past and future frames at different time offsets, thereby significantly increasing the diversity of motion information available for prediction without excessive complexity.
Data Source
AI summary
Embodiments of the present disclosure provide a solution for video processing. A method for video processing is proposed. The method comprises: determining, during a conversion between a target video block of a video and a bitstream of the video, at least one group of motion vector predictions (MVP) candidates of the target video block; determining an MVP candidate list by sorting the at least one group of MVP candidates based on respective template matching costs of MVP candidates in the at least one group; and performing the conversion based on the MVP candidate list. In this way, a proper MVP candidate list can be determined, and thus the coding effectiveness and coding efficiency can be improved.