A Novel Approach Extracting Extreme Points Combining Multidirection Phase Superposition and Weighted Moment Diagram for Matching Multi-Modal Images

Abstract

This paper proposed the novel approach of multi-model image matching based on the feature integration of Multi-directional Phase Superposition and weighted moment diagrams to handle the matching problem of large contrast differences, lots of noises and nonlinear radiometric distortion between an image pair. There are three steps for this method. Firstly, the presented method extracted the local extreme points in Multi-directional Phase Superposition diagram and the edge extreme points in the weighted moment diagram. Then, the characteristic descriptor was produced. Finally, angle-assisted difference measurement was put forward to handle the problem of angle reverse. In the collected data set, this method was experimentally compared with RIFT, CoFSM and HAPCG. The number of extreme points and Recall of PSMWD are higher than those of RIFT and HAPCG. In image pairs of noises, the number of correct matchings of PSMWD is about twice of HAPCG. Therefore, the proposed theory has practical importance.