Tighter Performance Bounds on Image Registration

Image registration is a fundamental and important task in image processing. It essentially estimates a transformation that aligns two images. Cramer Rao lower bound has recently been used to establish the performance limit of image registration algorithms. However, it is known to be a weak lower bound for some problems. In this paper, we analyze the mean square error performance of transformation estimation in image registration problems. We focus on rigid body transformations, and derive a set of tighter alternatives, namely the Bhattacharya bound and the Ziv-Zakai bound. Experimental results demonstrate the validity of our performance bounds