Video Encoding With Selective Bi-Directional Optical Flow
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video encoding technologies face inefficiencies due to increased processing requirements when performing subdivided prediction processes, leading to reduced coding efficiency.
Innovation Solution
An encoder and decoder that utilize a mode selection process for motion prediction, performing prediction based on motion vectors of each block or sub-block, with optional spatial gradient correction in one mode and omitting correction in another, to maintain coding efficiency while reducing processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a subdivided prediction process is performed for each block, then prediction accuracy is improved, but processing complexity increases
Solution Approach 1:
The current block is divided into multiple sub-blocks, and motion prediction is performed independently for each sub-block. This segmentation allows the prediction process to adapt to local motion variations within the block, improving prediction accuracy while managing processing complexity through structured subdivision.
Solution Approach 2:
The patent applies bi-directional optical flow prediction selectively - only when the prediction mode indicator signals that this mode should be used. This partial action approach avoids unnecessary processing overhead while maintaining high prediction accuracy when needed, resolving the contradiction between accuracy and complexity.
2Productivity
If bi-directional optical flow prediction is applied to all blocks, then coding efficiency is improved, but processing time increases
Solution Approach 1:
Bi-directional optical flow prediction is applied partially - only to blocks where the prediction mode indicator specifies this mode. This selective application maintains coding efficiency for blocks that benefit from it while avoiding unnecessary processing time consumption for blocks where simpler prediction methods suffice.
Solution Approach 2:
The prediction method is made dynamic through the prediction mode indicator, which allows the system to switch between different prediction approaches (bi-directional optical flow vs. other modes) based on block-specific characteristics. This dynamic adaptation optimizes the balance between coding efficiency and processing time.
Data Source
AI summary
An encoding method includes determining a mode to be applied to a current block, the mode including a sub-block mode and a first mode different from the sub-block mode, wherein in the first mode, inter-prediction parameters are inferred from a neighboring block neighboring the current block, wherein in the sub-block mode, the current block includes a plurality of sub-blocks, and inter-prediction parameters are provided for each of the plurality of sub-blocks; in response to the mode being determined to be the first mode, generating a prediction image for the current block by performing a bi-directional optical flow prediction process, wherein the bi-directional optical flow prediction process uses a spatial gradient for the current block; and in response to the mode being determined to be the sub-block mode, generating a prediction image for the current block based on an affine motion compensation by not performing the bi-directional optical flow prediction process.