Multivariate entropy detector based hybrid image registration

Image registration is used to match two images for spatial alignment and intensity alignment. One of the possible applications of image registration is for the evaluation of a printed image with respect to a given reference image. We propose a new hybrid image registration algorithm to identify the spatial or intensity variations between two color images. The proposed approach extracts salient descriptors from the two images using a multivariate entropy-based detector. The transformation parameters are obtained after establishing the correspondence between the salient descriptors of the two images, which yields better accuracy and lesser computational cost compared to the approaches present in the literature.