Image Registration Using Rotation Normalized Feature Points

Abstract

In this paper a registration method based on Harris corners is proposed. It is composed of three main steps: extraction of feature point using a Harris corner detector, obtaining the correspondence between the features points of the reference and input image based on rotation normalized gray level information around the corners, and estimating the transformation parameters mapping the input image to the reference one by applying RANSAC. Experimental results illustrate the proposed algorithm is robust to translation, rotation, and slight scaling.