Motion Estimation Block Matching Algorithm Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Motion estimation in video encoding is a time-consuming process, accounting for over half of the processing time in video encoding, which hinders encoding and communication performance.
Innovation Solution
An enhanced block matching technique is introduced that utilizes different motion vector predictors and multiple search patterns, including stationary and spatially proximate blocks, as well as spiral and gradient search techniques to accelerate the identification of matching reference blocks, reducing unnecessary search branches and processing time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional block matching is used for motion estimation, then motion estimation accuracy is maintained, but processing time increases significantly (accounting for over half of encoding time)
Solution Approach 1:
The patent applies preliminary action by using motion vectors from spatially adjacent blocks and temporally corresponding blocks as predictors before performing full block matching. This preliminary prediction step narrows the search space and reduces the number of candidate blocks that require exhaustive comparison, thereby maintaining motion estimation accuracy while significantly reducing processing time.
Solution Approach 2:
The patent segments the motion estimation process into multiple stages: first using simple predictor blocks (spatial and temporal), then progressively refining with more complex search patterns only for blocks that require higher precision. This segmentation allows the system to apply computationally intensive methods only where necessary, reducing overall processing time while maintaining accuracy for critical blocks.
2Reliability
If exhaustive block matching is performed to ensure accurate motion vector prediction, then encoding quality is improved, but encoding speed decreases
Solution Approach 1:
The patent applies partial action by performing exhaustive block matching only for selected blocks that require high precision, while using simplified predictor-based methods for other blocks. The system selectively applies full search patterns based on block characteristics, motion complexity, and precision requirements, thereby achieving good encoding quality for critical blocks while maintaining overall encoding speed.
Solution Approach 2:
The patent applies local quality by using different block matching strategies for different regions of the video frame. High-motion or high-importance regions receive more rigorous matching with multiple search patterns, while low-motion or less critical regions use simpler predictors. This localized approach ensures high reliability where needed while improving overall encoding productivity.
3Loss of time
If multiple search patterns are used to reduce search time, then processing speed increases, but algorithm complexity increases
Solution Approach 1:
The patent applies dynamics by making the search algorithm adaptive and flexible. Multiple search patterns (spiral, gradient, diamond) are available but dynamically selected based on block characteristics, motion magnitude, and precision requirements. The algorithm can switch between different search strategies during execution, optimizing for both speed and complexity management in different contexts.
Solution Approach 2:
The patent reduces algorithm complexity by performing preliminary classification of blocks to determine which search patterns are necessary. Simple predictor blocks are identified early and processed with minimal complexity, while only blocks requiring higher precision trigger more complex search patterns. This preliminary sorting reduces the average complexity across all blocks.
Data Source
AI summary
Techniques to perform fast motion estimation are described. An apparatus may comprise a motion estimator operative to receive as input a current frame and a reference frame from a digital video sequence. The motion estimator may generate and output a motion vector. The motion vector may represent a change in position between a current block of the current frame and a matching reference block of the reference frame. The motion estimator may utilize an enhanced block matching technique to perform block matching based on stationary and spatially proximate blocks. Other embodiments are described and claimed.