Motion Vector Pruning via Picture Order Count Comparison
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding techniques face inefficiencies due to redundant motion vector prediction candidates, which lead to resource wastage and reduced compression efficiency, particularly in video codecs like HEVC.
Innovation Solution
The implementation of a picture order count (POC) based pruning technique to eliminate redundant motion vector prediction candidates, enhancing the uniqueness and diversity of prediction candidates without requiring additional overhead, applicable to various video coding scenarios.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If motion vector prediction candidates are generated without pruning, then more prediction candidates are available for selection, but redundant candidates waste computational resources and reduce compression efficiency
Solution Approach 1:
The patent extracts and removes redundant motion vector prediction candidates from the candidate list by comparing POC values. When a candidate's reference picture POC matches an already-present candidate's reference picture POC, the duplicate candidate is pruned from the list, retaining only unique candidates while maintaining the full set of useful prediction options.
2Reliability
If all motion vector prediction candidates are retained, then prediction accuracy may be improved, but device complexity and processing overhead increase
Solution Approach 1:
The pruning process is performed automatically by the video coding device itself using the POC values already present in the motion vector candidates. The device compares POC values and self-prunes redundant candidates without requiring external intervention or additional complex processing, making the complexity management inherent to the coding process.
3Productivity
If motion vector prediction candidates are pruned based on POC values, then redundant candidates are removed improving efficiency, but the pruning accuracy must be sufficient to avoid removing useful candidates
Solution Approach 1:
The patent replaces complex redundancy detection mechanisms with a simpler POC value comparison approach. Instead of analyzing motion vector content, reference picture content, or other complex parameters to detect redundancy, the invention uses the already-available POC values as a proxy to identify and remove duplicate candidates, substituting a simple numerical comparison for more complex analysis.
Data Source
AI summary
In general, this disclosure described techniques for pruning a list of motion vector prediction candidates based on picture order count (POC) values. A video coding device may be configured to generate the list of motion vector prediction candidates for a prediction unit (PU) of video data. The video coding device may prune a first motion vector prediction candidate from the list of motion vector prediction candidates when a POC value for a reference picture identified by the first motion vector prediction candidate is equal to a second POC value of a reference picture identified by a second motion vector prediction candidate in the list of motion vector prediction candidates. The video coding device may code the PU using the pruned list.