Average Motion Vector Candidate Generation for Video Coding Parsing Errors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The High-Efficiency Video Coding (HEVC) standard faces challenges in parsing errors due to dynamic candidate set sizes in motion vector prediction, leading to potential parsing error propagation across pictures, which can be mitigated by using a fixed candidate set size but results in coding loss.
Innovation Solution
Generating an average candidate from original motion candidates in the candidate set, which includes averaging motion vectors without scaling, to include in the candidate set and determine a selected motion vector predictor for inter picture prediction modes like Inter, Merge, or Skip.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a dynamic candidate set size is used in motion vector prediction, then coding efficiency is improved, but parsing errors occur and propagate across pictures
Solution Approach 1:
The patent applies preliminary action by constructing the candidate set with a fixed size before motion vector prediction, ensuring that the candidate set structure is predetermined and stable. This prevents parsing errors during decoding while maintaining coding efficiency through the use of average motion vector candidates that are pre-calculated from spatial and temporal neighbors.
Solution Approach 2:
The patent changes the parameter of candidate set size from dynamic to fixed, while introducing average motion vector candidates as a new parameter type. This transformation stabilizes the candidate set structure for error-free parsing, while the average candidates maintain coding efficiency by providing additional motion prediction options.
2Reliability
If a fixed candidate set size is used in motion vector prediction, then parsing errors are prevented, but coding efficiency deteriorates
Solution Approach 1:
The patent merges spatial motion candidates and temporal motion candidates into a unified fixed-size candidate set. By combining these different types of motion candidates and calculating their averages, the patent creates a comprehensive candidate set that maintains coding efficiency while adhering to a fixed structure that prevents parsing errors.
Solution Approach 2:
The patent introduces average motion vector candidates as intermediaries between spatial and temporal motion candidates. These average candidates serve as mediators that provide additional motion prediction options within the fixed candidate set structure, thereby maintaining coding efficiency without compromising parsing reliability.
3Productivity
If average motion vector candidates are generated from spatial and temporal neighbors, then coding efficiency is maintained, but computational complexity increases
Solution Approach 1:
The patent applies local quality by calculating average motion vectors only for specific spatial and temporal neighbors that are most relevant to the current block. This selective averaging approach maintains coding efficiency while reducing unnecessary computational complexity compared to calculating averages for all possible neighbors.
Solution Approach 2:
The patent uses partial action by calculating average motion vectors for a limited set of spatial and temporal neighbors rather than all possible neighbors. This partial averaging provides sufficient coding efficiency improvement while keeping computational complexity manageable.
Data Source
AI summary
Video processing methods and apparatuses for coding a current block by constructing a candidate set including an average candidate generated from two or more original motion candidates. At least one MV of the average candidate is derived by directly averaging MVs of the original motion candidates in one of list 0 and list 1 without scaling regardless whether the MVs are pointing to same or different reference pictures. A selected candidate is determined from the candidate set as a MVP for a current MV of the current block for encoding or decoding the current block.