Inter-Frame Prediction Method Using BDOF Flag for Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing video coding and decoding processes using Bi-directional Optical Flow Prediction (BDOF) are complex and time-consuming due to the need for calculating spatial gradients, time gradients, auto-correlation parameters, cross-correlation parameters, optical flow vectors, and luma offset values for each sub-block, resulting in low calculation efficiency.
Innovation Solution
An inter prediction method that determines a forward and backward Motion Vector (MV) for a current block, sets a BDOF flag based on a set condition, and decides whether to perform BDOF on the block or sub-block, skipping unnecessary calculations for blocks or sub-blocks that do not satisfy a BDOF utilization condition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If BDOF is performed for each sub-block to obtain accurate luma prediction values, then prediction accuracy is improved, but calculation complexity and time consumption increase significantly
Solution Approach 1:
The current block is divided into multiple sub-blocks, and BDOF is selectively applied only to sub-blocks that meet specific conditions (e.g., gradient magnitude thresholds). This segmentation allows the system to maintain high prediction accuracy for complex regions while avoiding unnecessary calculations in simpler regions, thereby reducing overall calculation complexity.
Solution Approach 2:
Different processing strategies are applied to different sub-blocks based on their local characteristics. Sub-blocks with high gradient magnitudes (indicating complex motion or texture) undergo full BDOF processing, while sub-blocks with low gradient magnitudes use simplified processing. This local quality approach ensures prediction accuracy where needed while reducing computation elsewhere.
2Measurement precision
If BDOF calculations are performed for all sub-blocks, then prediction quality is maintained, but processing speed decreases
Solution Approach 1:
Gradient magnitude calculations are performed as a preliminary step before applying BDOF. By evaluating gradient magnitudes first, the system can identify sub-blocks that are likely to benefit from BDOF processing and skip calculations for sub-blocks that won't improve prediction quality significantly, thus maintaining prediction quality while improving processing speed.
Solution Approach 2:
Instead of applying BDOF to all sub-blocks (excessive action), the method applies BDOF only to a subset of sub-blocks that meet certain criteria (partial action). This selective application maintains adequate prediction quality for the overall block while significantly reducing the number of computations required, thereby improving processing speed.
3Measurement precision
If comprehensive gradient and correlation calculations are performed for BDOF, then prediction accuracy improves, but time consumption increases
Solution Approach 1:
The method extracts and evaluates key features (gradient magnitudes, motion vector differences) as preliminary criteria before committing to the full BDOF calculation process. By taking out these essential checks, the system can quickly identify candidate sub-blocks for BDOF processing and avoid the time-consuming full calculations for sub-blocks that don't meet the criteria, thus reducing overall time consumption while maintaining accuracy for relevant regions.
Data Source
Figure 1~2
Figure 3A
Figure 3B
AI summary
An inter-frame prediction method and apparatus, and related devices and a storage medium. The method comprises: determining a first motion vector and a second motion vector of a current block, wherein the first motion vector is a forward motion vector of the current block, and the second motion vector is a backward motion vector of the current block; on the basis of the first motion vector and the second motion vector, determining whether a set condition is satisfied, and obtaining a determination result; setting the value of a BDOF flag according to the determination result; and determining, on the basis of the value of the BDOF flag, an inter-frame prediction value of the current block or at least one sub-block in the current block.