On Image Registration using The Radon Transform: Review-and-Improvement

Abstract

In this paper, we review the problem of identifying a Linear Transformation applied on an image. Three major parts are presented, all involved the use of Radon Transform: First, recovering a sequence of basic transformations on an image; namely, reflection, rotation, dilation, and translation. Second, recovering a transformation on reference Image and an inspected image, where is obtained from by a general Linear Transformation. In doing so, we review our Analysis using the Singular Value Decomposition of the Transformation's Matrix. Third, we present an alternative efficient method of obtaining a matrix of transformation by testing a well-defined class of potential matrices using only the two projections of the inspected image.