Improvement of Modal Matching Image Objects in Dynamic Objects Matching Improvement using Optimization Techniques in a Geometric Modal Methodology

Abstract

This work was based on a previously developed methodology using shape modal description for matching image objects. With this methodology, the objects points were matched using a local search approach. This paper presents a global search methodology using optimization techniques to establish the correspondences between two objects. It was also developed a solution to match excess points after the usual "one to one" matching, if the number of nodes that compose both objects is not the same. It is also briefly presented the method used for the determination of the rigid transformation between two objects, which facilitates the results analysis.