Bi-prediction Coding Motion Vector Scaling and Cost Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional bidirectional prediction coding methods face challenges in complexity and coding efficiency, particularly when dealing with linearly moved images, as they often transmit only the forward motion vector, leading to increased bit rates and reduced compression efficiency.
Innovation Solution
A bi-prediction coding method that selects and codes motion vectors based on multiple reference pictures, calculating and comparing coding costs to determine the most efficient motion vectors for coding and non-coding targets, thereby reducing bit rates and enhancing coding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If joint motion search is performed to maximize motion vector accuracy, then prediction error is minimized, but computational complexity increases significantly
Solution Approach 1:
The patent segments the bidirectional prediction process into two independent unidirectional prediction processes. Instead of performing joint motion search for both forward and backward motion vectors simultaneously, the patent performs separate motion searches for each direction, reducing computational complexity while maintaining prediction accuracy through subsequent weighted combination of the two prediction results.
Solution Approach 2:
The patent merges the results of two independent unidirectional prediction processes through weighted combination. By combining the forward prediction block and backward prediction block with appropriate weights, the patent achieves bidirectional prediction效果 without the high complexity of joint motion search, thus resolving the contradiction between accuracy and complexity.
2Measurement precision
If both forward and backward motion vectors are transmitted for bidirectional prediction, then prediction accuracy is improved, but bit rate increases
Solution Approach 1:
The patent extracts only the essential forward motion vector information for transmission, and derives the backward motion vector information through calculation based on the forward motion vector and temporal relationships. This extraction approach reduces the amount of data that needs to be transmitted while maintaining the ability to perform accurate bidirectional prediction at the decoder side.
3Device complexity
If independent forward and backward motion vector search is performed, then computational complexity is reduced, but prediction error increases
Solution Approach 1:
The patent implements a feedback mechanism where the forward motion vector search results are used to inform and optimize the backward motion vector search. By using the forward motion vector information as feedback, the backward search can be more targeted and efficient, reducing the overall prediction error while maintaining lower computational complexity compared to fully independent searches.
Data Source
AI summary
An image decoding method and an image decoding apparatus is provided. The method comprises recovering a first motion vector corresponding to a first decoding reference picture based on the entropy decoded bit stream, calculating a second motion vector corresponding to a second decoding reference picture by scaling the first motion vector based on a first temporal distance between the current picture and the first decoding reference picture and a second temporal distance between the current picture and the second decoding reference picture, generating a prediction block relating to a current block in the current picture, based on the calculated second motion vector, generating a residual block relating to the current block through a residual data decoding process based on the entropy decoded bit stream, and recovering the current block based on the prediction block and the residual block.