Motion Vector Prediction Across Multiple Resolutions for Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video encoding and decoding methods face inefficiencies in determining optimal motion vector predictors and resolutions, leading to increased complexity and reduced compression rates.
Innovation Solution
The method involves adaptively determining motion vector predictors and resolutions by using spatial and temporal candidate blocks, encoding information representing the motion vector predictor, residual motion vector, and motion vector resolution, and employing interpolation techniques to improve encoding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If motion vector predictors are determined using multiple resolutions and candidate blocks, then prediction accuracy is improved, but device complexity increases
Solution Approach 1:
The patent applies dynamics by adaptively selecting motion vector resolution based on block characteristics and motion complexity. The system dynamically adjusts between different resolution levels (e.g., 1/4-pel, 1/8-pel, 1/16-pel) rather than using a fixed high resolution for all blocks, thereby improving prediction accuracy where needed while reducing overall computational complexity.
Solution Approach 2:
The patent implements local quality by applying different motion vector resolutions to different blocks based on their specific characteristics. High-resolution motion vectors are used for blocks with complex motion or high importance, while lower resolutions are used for simpler blocks, optimizing the balance between prediction accuracy and complexity on a block-by-block basis.
2Measurement precision
If high resolution motion vectors are used for all blocks, then prediction precision is improved, but compression rate decreases
Solution Approach 1:
The patent applies parameter changes by varying the motion vector resolution parameter according to block characteristics, motion complexity, and importance. The encoder selectively uses different resolution levels (1/4-pel, 1/8-pel, 1/16-pel) to optimize the balance between prediction precision and bitrate consumption, rather than uniformly applying high resolution to all blocks.
Solution Approach 2:
The patent implements partial action by applying high-resolution motion vectors only to specific blocks that require such precision, rather than to all blocks. This selective approach ensures that the additional precision is obtained only where necessary, maintaining compression efficiency while improving prediction accuracy for critical blocks.
3Measurement precision
If multiple motion vector predictor candidates are evaluated, then optimal prediction is achieved, but processing time increases
Solution Approach 1:
The patent applies segmentation by dividing the motion vector prediction process into stages with different numbers of candidates evaluated at each stage. The encoder generates multiple candidate blocks and motion vector predictors, but evaluates them in a structured manner that allows early termination or selective processing, reducing overall processing time while still achieving optimal prediction.
Solution Approach 2:
The patent implements dynamics by adaptively adjusting the number of motion vector predictor candidates evaluated based on block characteristics and available processing resources. The system can dynamically reduce the candidate evaluation process for blocks with simple characteristics while maintaining thorough evaluation for complex blocks, optimizing the trade-off between prediction optimality and processing time.
Data Source
Figure 1A~1B
Figure 2A~2B
Figure 3A
AI summary
A motion vector encoding apparatus includes: a predictor configured to obtain motion vector predictor candidates of a plurality of predetermined motion vector resolutions by using a spatial candidate block and a temporal candidate block of a current block, and to determine motion vector predictor of the current block, a motion vector of the current block, and a motion vector resolution of the current block by using the motion vector predictor candidates; and an encoder configured to encode information representing the motion vector predictor of the current block, a residual motion vector between the motion vector of the current block and the motion vector predictor of the current block, and information representing the motion vector resolution of the current block, wherein the plurality of predetermined motion vector resolutions include a resolution of a pixel unit that is greater than a resolution of one-pel unit.