Bi-Predictive Motion Vectors in GPM Subblocks for Video Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face challenges in efficiently utilizing bi-predictive motion vectors and geometric partition modes to enhance compression efficiency, particularly in handling bi-directional optical flow and decoder-side motion vector refinements.
Innovation Solution
Implementing a method and apparatus for video encoding/decoding that utilize geometric partition modes (GPM) with bi-predictive motion vectors, dividing blocks into larger subblocks of N×N size and applying bi-directional optical flow (BDOF) or decoder-side motion vector refinement (DMVR) based on blending masks to refine motion vectors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If subblock based motion refinement (BDOF/DMVR) is applied to all subblocks in GPM mode, then motion vector refinement quality is improved, but computational complexity and processing time increase significantly
Solution Approach 1:
The patent applies different subblock sizes (N×N where N≥4) for different regions within the current block based on blending mask values. Regions with higher blending mask values (indicating stronger motion gradients) receive finer subblock partitioning, while regions with lower values use coarser partitioning. This local adaptation of subblock granularity optimizes motion refinement quality where needed while reducing computational load in regions requiring less precision.
Solution Approach 2:
The patent divides the current block into multiple N×N subblocks before applying motion refinement, where N is at least 4. This segmentation allows the motion refinement process to operate on smaller, more manageable units with bi-predictive motion vectors, improving both the quality and efficiency of the refinement process by processing localized motion characteristics independently.
2Productivity
If larger N×N subblock sizes are used for motion refinement, then processing speed is improved, but motion compensation precision deteriorates
Solution Approach 1:
The patent dynamically selects the subblock size N (where N≥4) based on the characteristics of the current block, particularly the blending mask values and motion complexity. This dynamic adaptation allows the system to use larger subblock sizes for simple regions (improving speed) while using smaller subblock sizes for complex regions (maintaining precision), thereby resolving the trade-off between processing speed and motion compensation precision.
Solution Approach 2:
Different N×N subblock sizes are applied to different regions of the current block based on local motion characteristics indicated by blending mask values. Regions with simple motion patterns use larger N values for faster processing, while regions with complex motion patterns use smaller N values for higher precision, achieving both speed and accuracy in different locations simultaneously.
3Productivity
If bi-predictive motion vectors are used in GPM mode, then compression efficiency is improved, but decoder complexity increases
Solution Approach 1:
The patent applies bi-predictive motion vectors to N×N subblocks (where N≥4) within the current block, dividing the prediction task into smaller independent units. Each subblock has its own bi-predictive motion vector derived from reference blocks, allowing the decoder to process motion compensation in a modular fashion that improves compression efficiency while managing complexity through localized processing.
Solution Approach 2:
The patent changes the granularity parameter N (where N≥4) to control the number and size of subblocks receiving bi-predictive motion refinement. By adjusting N, the system can control the number of bi-predictive motion vectors that need to be decoded and processed, thereby managing decoder complexity while maintaining compression efficiency through adaptive subblock partitioning.
Data Source
AI summary
Some aspects of the disclosure provide an apparatus for video decoding. The apparatus includes processing circuitry configured to receive a coded video bitstream comprising coded information of one or more pictures, determine, from the coded information, that a current block in a current picture is in a geometric partition mode (GPM) with at least a first GPM partition having a bi-predictive motion vector, and apply subblock based motion refinements with bi-directional motion on at least a first subblock and a second subblock of the first GPM partition. The first subblock and the second subblock have different subblock sizes. The processing circuitry reconstructs the current block based on the subblock based motion refinements.