Research Of Image Registration Based On Mutual Information And Particle Swarm Algorithm

The image registration technology has become a research hotspot in the field of computer at present, and it is an important task for the image registration that comprehensive analysis the image information which got from different sensors. Image registration technique is an important prerequisite for other image processing fields such as image fusion, and the result of mage registration can impact directly on image fusion. In medicine, the registration of the medical image which made by different sensors in different time can get more comprehensive information of the patients to diagnose and treat.In this article, the task of image registration is the using an efficient optimization algorithm to find a set of optimal space transformation parameter which achieve the maximum similarity measure. This paper first introduces the basic knowledge of image registration which serves as the image registration algorithm design theory. The mutual information is selected for the similarity measure. This paper studied the application of image registration which is based on the particle swarm algorithm according to the analysis of the advantages and disadvantages of global algorithm and local algorithm.Aim at the defects that article swarm algorithm easy to fall into local extremum leading to the low registration precision, this paper design a hybrid pso algorithm based on ant colony algorithm. In the process of registration, solve the optimal space transformation parameters which achieving the maximum mutual information value by the hybrid algorithm. The experiments proved that the Registration accuracy based on the improved algorithm is higher than that based on particle swarm optimization (pso) algorithm.In addition, in the late stage of image registration, aim to the defects that particle of the particle swarm optimization (pso) algorithm and hybrid pso algorithm based on ant colony algorithm will incline to identity gradually and lose diversity, this paper introduces the Wild goose team theory and designs the hybrid pso algorithm based on Wild goose team theory and ant colony algorithm. Experiments show that the registration precision based on the hybrid algorithm is higher than the first two algorithms registration precision.The two improved algorithms in this paper will provide theoretical directions and practical guidance for the deployment of image registration in medical.