C15. Registration of multi-focus images using Hough transform

Image registration is the process of finding a transformation that aligns one image to another. Most researchers who are interested in the registration task don't take into consideration the out-of-focus phenomenon that might affect the accuracy of the registration process. Nevertheless, they suggest that the images are well focused. Consequently, this work deals with the registration of multi-focus images; the out of focus is being treated side by side with the registration process. This paper presents an accurate and fast method for rigid registration of multi focus images. It is based on Hough transformation of images followed by Particle Swarm Optimizer (PSO) as an optimization technique and mutual information (MI) as a cost function (similarity measure). Hough transform introduces a proper initialization for the optimizer. Maximizing the MI between reference and sensed images will lead to minimizing the misalignment error between reference and sensed images, hence yields to accurate registration.