AUTOMATED IMAGE REGISTRATION BY MAXIMIZATION OF A REGION SIMILARITY METRIC

This article presents a robust algorithm for automated registration of images related by rigid-body transformations. This algorithm uses a n ew region-based similarity metric, which enables accurate registration of images of large contrast differences. Region segmentation required by the metric is accomplished using a multiscale segmentation algorit hm, and minimization of this metric is done using the Powell direction set method. Experimental results are presented to demonstrate that th e algorithm is effective for aligning images from single or multiple i maging modalities without the use of any fiducial markers. (C) 1997 Jo hn Wiley & Sons, Inc.