Evaluating Robustness for Intensity Based Image Registration Measures Using Mutual Information and Normalized Mutual Information

Abstract

Numerous Applications envisage the need of image registration. The term image registration is the process of alignment of multimodal, multitemporal or multiview images onto a single co-ordinate system by applying certain transformations e.g. rotation, translation, scaling etc. This paper uses Information Measures such as Mutual Information (MI), and Normalized Mutual Information (NMI) to obtain the aligned image and then evaluate their robustness. Various Parameters such as Pearsons correlation coefficient (PCC), Peek Signal to Noise Ratio (PSNR), have been used to quantify the image registration results obtained through MI and NMI. The methodology for registration has been validated using above parameters and elapsed time is also evaluated for rigid registration of natural Images.