Bi-Directional Optical Flow Prediction With Derived Motion Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image encoding/decoding methods using bi-directional optical flow (BIO) are limited when only one piece of motion information is available, leading to inefficiencies in high-resolution and high-quality image data transmission and storage.
Innovation Solution
A method and apparatus that derive second motion information when a block includes only one piece of motion information, allowing bi-directional prediction, with variable sub-group size for calculating BIO offsets and the option to apply BIO in a block unit to reduce complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If bi-directional optical flow prediction is applied requiring two pieces of motion information, then prediction accuracy is improved, but applicability deteriorates when only one piece of motion information is available
Solution Approach 1:
The patent applies preliminary action by deriving second motion information in advance when only one piece of motion information is available. The system proactively creates the necessary second motion vector by reversing the first motion vector, enabling bi-directional prediction to proceed without requiring actual second motion measurements or data.
Solution Approach 2:
The patent uses an intermediary approach by introducing derived motion information as a mediator between the available first motion information and the bi-directional prediction process. This derived second motion information acts as a placeholder or substitute that enables the prediction system to function with incomplete data.
2Productivity
If bi-directional optical flow prediction is applied to derive second motion information, then coding efficiency is improved, but computational complexity increases
Solution Approach 1:
The patent applies partial action by implementing bi-directional optical flow prediction selectively rather than universally. The system determines whether to derive second motion information based on specific conditions (such as motion vector magnitude thresholds), applying the complex BIO process only when beneficial and using simpler unidirectional prediction when sufficient.
Solution Approach 2:
The patent changes parameters to control the complexity of the prediction process. By adjusting thresholds for motion vector magnitudes and other parameters, the system can switch between deriving second motion information (higher complexity) and using only first motion information (lower complexity), thereby adapting computational requirements to coding efficiency needs.
3Ease of manufacture
If fixed sub-group size is used for calculating BIO offset, then calculation is simplified, but flexibility deteriorates for different block sizes
Solution Approach 1:
The patent applies dynamics by making the sub-group size variable rather than fixed. The system dynamically adjusts the sub-group size based on the block size and other parameters, allowing the BIO offset calculation to adapt to different image regions and prediction scenarios while maintaining computational feasibility through structured division of the block.
Data Source
AI summary
An image encoding/decoding method is disclosed. An image decoding method of the present invention may comprises determining step for determining whether or not to derive second motion information when a current block includes first motion information only and for which bi-directional prediction is available, a deriving step for deriving the second motion information based on the first motion information, and a predicting step for generating a prediction block of the current block by performing BIO prediction based on the first motion information and the second motion information.