Points matching via iterative convex hull vertices pairing

Image registration is a fundamental task in image processing. In feature-based image registration, feature matching is the key but difficult step. In this paper, a point matching algorithm, which takes use of convex hull of point set and establishes correspondence between matched hull vertices, is proposed. Mapping parameters and point sets are updated after each step. After all points are processed, the current mapping parameter is the exact parameter between images. Experiment indicated that this method can match two point sets correctly with a little manual interference.