Motion Vector Candidate Selection for Efficient Video Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video encoding formats like H.264/AVC suffer from insufficient motion vector prediction accuracy, leading to a large number of coded bits.
Innovation Solution
A method for dynamically selecting a predictive motion vector based on the distribution of candidate vectors, using thresholds to determine the complexity of motion and adjusting the code table selection to match the image characteristics, thereby reducing the amount of coded bits.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional motion vector prediction methods are used in H.264/AVC format, then the encoding process is simple, but the prediction accuracy is insufficient and the amount of coded bits for motion vectors remains large
Solution Approach 1:
The patent applies dynamics by making the prediction method adaptive rather than static. The encoder dynamically selects between two different prediction approaches (conventional median-based prediction and new candidate-based prediction) based on the actual motion characteristics of the video content. This allows the system to optimize prediction accuracy for each specific block while maintaining efficiency through conditional selection rather than always using the more complex method
Solution Approach 2:
The patent changes the prediction parameter by introducing multiple candidate motion vectors (left, above, above-right candidates) and using deviation evaluation metrics to select the best prediction method. Instead of always using the simple median of three candidates, the system evaluates the distribution characteristics of these candidates and switches between prediction strategies based on measured deviation values, thereby adapting to different motion patterns
2Productivity
If the amount of coded bits for motion vectors is reduced through prediction, then compression efficiency improves, but the complexity of the encoding process increases
Solution Approach 1:
The patent segments the prediction process into distinct evaluation stages: calculating deviation metrics for each candidate block, comparing these deviations against thresholds, and selectively applying different prediction methods based on the evaluation results. This segmented approach with clear decision boundaries (thresholds) manages complexity by breaking down the adaptive selection process into manageable, conditional steps rather than a monolithic complex algorithm
Solution Approach 2:
The patent implements feedback through deviation evaluation, where the encoder measures the actual distribution characteristics of candidate motion vectors and uses this feedback to determine the optimal prediction method. The deviation metrics provide quantitative feedback about the suitability of different prediction approaches, enabling the system to adaptively optimize compression efficiency while maintaining controlled complexity through objective decision criteria
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A video decoding method in which a motion vector for a block to be decoded is calculated based on information decoded from a coded stream, by executing motion vector prediction and by adding a differential motion vector read out from the coded stream to a predicted motion vector, and an inter-prediction process is performed using the motion vector, wherein in the motion vector calculation, three motion vector values of three decoded blocks including decoded blocks located near the block to be decoded, which are a first, second and third motion vector value, are examined as options for the predicted motion vector. If a difference between two motion vector values among the first to third motion vector values, is smaller than a threshold value, only one motion vector among the two motion vectors remains as one of options for the predicted motion vector, then the remaining one motion vector value among the two motion vector values and the other one motion vector value than the two motion vector values among the three motion vector values remain as two options for determining the motion vector value of the predicted motion vector. A motion vector list listing the two options of the motion vector values is generated and selection information is retrieved from the coded stream, such that the predicted motion vector is determined based on the selection information among the motion vector values listed in the motion vector list.