Downhill Simplex Motion Estimation for Video Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing motion estimation algorithms for video compression, such as full search, have high computational costs, making them impractical for real-time applications, and fast block matching algorithms often compromise on video quality by reducing the number of search points.
Innovation Solution
The proposed method uses a downhill simplex search algorithm to iteratively refine motion vectors by replacing points with lower function values, selecting candidates from mean values of preceding and succeeding blocks, and zero vectors, and extends this to multi-reference frames by forming an initial simplex from motion vectors across multiple frames.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If full search algorithm is used for motion estimation, then motion vector accuracy is improved, but computational cost increases significantly
Solution Approach 1:
The patent applies preliminary action by using motion vectors from adjacent blocks and temporal prediction from previous frames to pre-determine the initial simplex vertices. This preliminary estimation narrows down the search space before the actual optimization begins, allowing the algorithm to achieve high accuracy without exhaustively searching all possible locations, thus reducing computational cost while maintaining motion vector precision
Solution Approach 2:
The patent replaces the mechanical exhaustive search system with a mathematical optimization system based on simplex algorithm. Instead of systematically checking every possible motion vector location, the system uses function evaluation and geometric transformation (reflection, expansion, contraction, shrinkage) to converge to the optimal motion vector, significantly reducing computational complexity while maintaining accuracy
2Device complexity
If fast block matching algorithms are used to reduce search points, then computational complexity is reduced, but video quality degrades
Solution Approach 1:
The patent replaces traditional fast block matching algorithms with a downhill simplex search algorithm that uses continuous function evaluation and geometric optimization. This substitution allows the system to achieve high precision motion estimation with fewer search points by using mathematical optimization rather than discrete block comparison, thereby maintaining video quality while reducing computational complexity
Solution Approach 2:
The patent changes the optimization parameters by using sum of squared errors (SSE) as the objective function and employing adaptive simplex operations (reflection factor, expansion factor, contraction factor). These parameter changes enable the algorithm to converge to accurate motion vectors with fewer iterations compared to traditional fast BMAs, thus maintaining video quality while reducing the number of search points required
Data Source
AI summary
A motion estimation method for video compression comprises the following steps. First, an initial simplex comprising three points is determined based on motion vectors in blocks of a current frame and a previous frame, and a point having a largest function value among the three points is replaced with a point having a smaller function value to form a simplex. The replacement is repeated until two points of the three points of the simplex converge to a same point. The iteration is performed by downhill simplex search including operations of reflection, expansion, contraction and shrinkage to find a point for replacement. The motion estimation method for video compression can also use multi-reference frames. An initial simplex comprising four points is determined based on motion vectors of a current frame with reference to a plurality of previous frames, and a point having a largest function value among the four points is repeatedly replaced with a point having a smaller function value to form a simplex until two points of the four points of the simplex converge to a same point.