A feature-based image registration technique for images of different scale

In this paper a technique for image registration of two geometrically distorted images is presented. These images may be further degraded by noise, blurring, etc., and may have only partial overlap. The geometric distortions considered in the registration process are the global 2D affine transformations including scaling and shearing. The proposed technique consists of three main steps: extracting feature point using a feature point extractor based on scale-interaction of Mexican-hat wavelets, obtaining the correspondence between the feature points of the reference and the target images based on Zernike moments of neighborhoods centered on the feature points, and estimating the transformation parameters between the first and the second images using an iterative weighted least squares algorithm. Experimental results illustrate the accuracy of image registration of images with various geometric distortions including different scales in the presence of additional image degradations.