Motion Vector Candidate Restriction for CTU-Based Video Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video compression technologies face challenges in balancing compression efficiency and computational complexity, particularly in motion vector prediction, as they rely on a large set of potential prediction candidates that can lead to inefficient data access and increased memory requirements.
Innovation Solution
The method involves generating a list of motion vector candidates based on relative positions within a coding tree unit, adjusting and replacing positions to ensure only available motion information is used, thereby reducing unnecessary data access and memory usage while maintaining prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a large set of potential prediction candidates is used in motion vector prediction, then prediction accuracy is improved, but computational complexity and memory requirements increase
Solution Approach 1:
The patent segments the motion vector prediction process into two distinct modes: merge mode with a limited set of 6 candidates and advanced motion vector prediction mode with up to 12 candidates. This segmentation allows the system to choose the appropriate complexity level based on prediction needs, resolving the contradiction between accuracy and computational load by providing graded options rather than a single large candidate set.
Solution Approach 2:
The patent implements partial action by providing a smaller subset of candidates (merge mode with 6 candidates) for cases where full prediction accuracy is not required, and excessive action by providing a larger set (AMVP mode with 12 candidates) when higher accuracy is needed. This allows the system to apply just the right amount of computational effort for each specific prediction scenario.
2Measurement precision
If a large set of potential prediction candidates is used in motion vector prediction, then prediction accuracy is improved, but memory requirements increase
Solution Approach 1:
The patent segments the candidate sets into two groups: a small merge mode candidate list (6 candidates) and a larger AMVP candidate list (12 candidates). By maintaining separate, appropriately-sized candidate lists for different prediction modes, the patent reduces overall memory requirements compared to always storing a large candidate set, while still providing access to comprehensive candidates when needed.
Solution Approach 2:
The patent applies local quality by having different candidate set sizes for different prediction modes and different spatial locations (merge mode for some blocks, AMVP for others). This allows memory to be optimized locally for each prediction scenario rather than uniformly allocating memory for the largest possible candidate set throughout the entire video processing system.
3Measurement precision
If comprehensive motion vector prediction candidates are considered, then prediction quality is improved, but data access efficiency decreases
Solution Approach 1:
The patent segments the data access requirements by creating two separate candidate generation processes: merge mode that accesses a limited set of neighboring block motion vectors, and AMVP mode that accesses additional temporal and spatial candidates. This segmentation improves data access efficiency by avoiding the need to always access the full set of comprehensive candidates, while still providing access to comprehensive data when prediction quality requires it.
Data Source
AI summary
The present disclosure relates to encoding and decoding of video images. Motion vectors are used as predictors for prediction of an image for a current coding unit CU, which is within a current coding tree unit CTU. The motion vectors are taken from a list of motion vector candidates MVCs. The list is generated for the current CU using a pattern of positions of allowed motion vectors, and in dependence on the positions of the pattern relative to the current CU, and on the position of the current CU within the current CTU. From the pattern, a subset of positions is selected and positions are restricted to positions within the current CTU, while other pattern positions not available or not stored in a buffer are excluded. Other excluded pattern positions may be substituted and/or adjusted to as to become available and/or are being stored in the buffer.