A Novel Phase Congruency-Based Image Matching Method for Heterogenous Images

Since different imaging principles of different ges lead to imaging results exhibiting nonlinear intensity differences, this phenomenon makes the traditional image alignment methods based on image gradients challenging. To solve this problem, instead of using image pixel intensity as the basis for image features, this paper searches for the relationship between phase and contour structure in digital images and summarizes this relationship as a new method for describing image structure. The method is light and contrast invariant, and furthermore, after phase voting, the method has a more powerful description between structure spaces. The proposed method is effective for image alignment of images from different sources.