Image registration, parameter tuning and approximate function evaluation, using the genetic algorithm and digital image warping

A common problem in image processing is image registration. The problem is finding the parameters of a noisy, multi-modal, multi-dimensional, image warping function which aligns two images. An image registration program, which used a genetic algorithm as the function optimiser, was designed. The settings of the program were automatically tuned with a second genetic algorithm. Monte Carlo analysis was used to improve the results.