Image registration based on weighted residual complexity using local entropy

Abstract

In this paper, we propose a novel intensity-based similarity measure to match medical images with intensity distortions. A specifically designed weighting function was incorporated into the residual term in residual complexity to weight the residual image between reference image and warped floating image adaptively. The weighting function is modeled by a function of the local entropy dierence after mapping images to local entropy space respectively. The experimental results show that our proposed approach has achieved more accurate and robust performance than mutual information and residual complexity.