Image Registration Algorithm Based on Copula Distribution Estimation

Abstract

In order to solve the problem that the measure function is easy to fall into the local extremum because of much local extremes in the mutual information registration method, this paper designs an optimization algorithm applied to image registration. One measure function based on mutual information and gradient similarity is constructed combining with evolutional copula estimation of distribution algorithm (ECEDA). New algorithm fully pays attention to the correlation between multidimensional variables joining with the linear weighted based on nonparametric estimation method to overcome the randomness of nonparametric estimation method. Experimental results show that comparing with the traditional copula estimation of distribution algorithm and other existing registration algorithms, the proposed algorithm has higher accurate rate and robustness.