History-Based Affine Motion Vector Prediction Buffer Management
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding standards face challenges in efficiently managing bandwidth and harmonizing affine prediction with other coding tools, particularly in handling complex motions like zoom, rotation, and perspective within video frames.
Innovation Solution
The implementation of geometry partitions with history-based motion vector prediction (HMVP) using 4-parameter and 6-parameter affine models, which involves storing and processing motion vector parameters to derive affine merge candidates and advanced motion vector prediction candidates, allowing for more flexible bandwidth management and improved prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If affine prediction with history-based motion vector prediction is implemented, then prediction accuracy for complex motions is improved, but device complexity and buffer management complexity increase
Solution Approach 1:
The patent segments the motion vector prediction process by maintaining a separate history buffer specifically for affine motion parameters. This buffer stores previously decoded affine motion vectors and parameters, which are then used to predict motion vectors for current blocks. The segmentation isolates the complex affine prediction logic from the general motion compensation process, improving accuracy for complex motions while managing device complexity through modular buffer management.
Solution Approach 2:
The patent implements preliminary action by pre-storing affine motion parameters from previously decoded blocks in a history buffer before they are needed for prediction. This allows the decoder to quickly access and utilize historical affine motion information without performing complex calculations in real-time, thereby improving prediction accuracy while keeping the processing complexity manageable through advance preparation.
2Measurement precision
If parameters for multiple video blocks are stored in buffer, then prediction accuracy is improved, but buffer management complexity increases
Solution Approach 1:
The patent applies parameter changes by dynamically adjusting the buffer management strategy based on the type of motion parameters being stored. Different affine motion parameters (such as 4-parameter vs. 6-parameter affine models) are handled with appropriate buffer sizes and retention policies. This allows the system to maintain high prediction accuracy by storing necessary historical parameters while managing buffer complexity through adaptive parameter-based management rather than uniform handling of all motion data.
Data Source
AI summary
Sub-table for history-based affine mode is described. In an exemplary aspect, a method for video processing includes deriving, deriving, for a conversion between a current block of video and a bitstream representation of the current block, an affine model for the current block based on affine related information, wherein the affine related information is stored in at least one first affine history motion vector prediction (HMVP) buffer of M first affine HMVP buffers or at least one affine sub-buffer of M affine sub-buffers (HB0, HB1, . . . HBM-1) into which a second affine HMVP buffer is divided, wherein the first affine HMVP buffer or affine HMVP sub-buffer being used to store affine related information of blocks coded prior the current block, M being an integer larger than 1; and performing the conversion by using the derived affine model.