GPM Optical Flow Refinement for Bi-Predictive Motion Blocks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face inefficiencies in compressing video data due to limitations in handling spatial and temporal redundancies, particularly in geometric partition modes with bi-predictive motion vectors, leading to suboptimal compression and decoding performance.
Innovation Solution
Implementing a method and apparatus for video encoding and decoding that utilize a sample-based bi-directional optical flow (S-BDOF) motion refinement, which applies spatial and subblock-based refinements based on geometric partition mode (GPM) partitions with bi-predictive motion vectors, and employs flags and thresholds to determine the application of S-BDOF motion refinement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional video coding techniques are used for geometric partition modes with bi-predictive motion vectors, then the coding complexity is reduced, but the compression efficiency and reconstruction quality deteriorate
Solution Approach 1:
The patent applies different refinement strategies to different regions within a partition. Sample-based BDOF is applied to specific samples (e.g., boundary samples or selected interior samples) rather than all samples, while subblock-based BDOF is applied to specific subblocks. This localized approach improves compression efficiency by focusing computational resources on regions that benefit most from refinement, rather than uniformly processing the entire block.
Solution Approach 2:
The patent divides a partition into multiple subblocks and applies motion refinement selectively to certain subblocks based on criteria such as subblock size, position, or motion characteristics. This segmentation allows the system to reduce complexity by skipping refinement in regions where it provides minimal benefit, while maintaining high compression efficiency in regions where refinement is most effective.
2Measurement precision
If sample-based BDOF motion refinement is applied to all samples in GPM partitions, then the reconstruction quality is improved, but the computational complexity increases
Solution Approach 1:
Instead of applying sample-based BDOF to all samples in a partition, the patent applies it only to a subset of samples that are deemed most beneficial for reconstruction quality. This could include boundary samples, samples in regions with high motion variation, or samples selected based on gradient criteria. This partial application maintains motion estimation accuracy where it matters most while significantly reducing computational complexity.
Solution Approach 2:
The patent introduces control parameters such as flags to enable or disable sample-based BDOF for different partitions or subblocks, and thresholds to determine which samples or regions qualify for refinement. These parameters allow dynamic adjustment of the refinement scope based on content characteristics, balancing accuracy and computational load adaptively.
3Reliability
If subblock-based BDOF refinement is applied to all subblocks, then the motion compensation accuracy is improved, but the processing time increases
Solution Approach 1:
The patent applies subblock-based BDOF refinement selectively to specific subblocks within a partition rather than uniformly to all subblocks. Selection criteria may include subblock size (e.g., only applying to larger subblocks), position within the partition, or local motion characteristics. This localized refinement improves motion compensation accuracy in critical regions while reducing processing time by skipping refinement in regions where it provides minimal benefit.
4Productivity
If adaptive selection of S-BDOF application is implemented, then the compression performance is optimized, but the decision-making complexity increases
Solution Approach 1:
The patent employs control parameters such as syntax flags and thresholds that govern when and where sample-based BDOF is applied. These parameters can be set based on partition characteristics, motion characteristics, or encoder/decoder capabilities. The decision logic, while adaptive, relies on comparing simple metrics (e.g., gradient magnitudes, motion vector differences) against thresholds, keeping the decision-making process relatively simple despite the adaptive nature.
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), and determine whether to apply a sample based bi-directional optical flow (S-BDOF) motion refinement on at least a first GPM partition of the current block. The first GPM partition has a bi-predictive motion vector. The processing circuitry is further configured to apply the S-BDOF motion refinement on one or more samples in the first GPM partition when applying the S-BDOF motion refinement on the first GPM partition is determined, and reconstruct the current block with the one or more samples being reconstructed based on the S-BDOF motion refinement.