Image Registration using Bio-inspired Algorithms

Abstract

Image registration is one of the most essential applications of image processing. In image registration, two images are compared to find a similarity metric and necessary adjustments are made to one of the images to minimize the similarity metric and align it to the other one (reference image). This minimization is performed using an optimization algorithm. Here, some of the newly developed meta-heuristic algorithms, namely Bat Algorithm and Grey Wolf Optimization are used to implement the image registration process with Mutual Information as the similarity metric. Along with these a Particle Swarm Optimization based image registration is also performed to the same sample sets. The performance results of these three implementations are compared on basis of both speed and quality of registration to find the overall best solution. The three algorithms are found to be very competitive when compared as optimizer in image registration process.