改进粒子群算法与多分辨率结合的医学图像配准  

Medical Image Registration Based on Improved Particle Swarm Optimization Algorithm and Multi-resolution

作　　者：王栋栋 宋余庆[1] 刘哲[1,2] 

机构地区：[1]江苏大学计算机科学与通信工程学院,江苏镇江212013 [2]吉林师范大学计算机学院,吉林四平136000

出　　处：《小型微型计算机系统》2017年第11期2610-2614,共5页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(61402204;61572239)资助;江苏省自然科学基金项目(BK20130529)资助

摘　　要：基于粒子群算法和以互信息为测度函数的医学图像配准方法拥有准确度高,健壮性强的优点,然而也会有局部收敛等不足.为了解决这些不足,将自我调节策略引入粒子群算法后与多分辨率相结合的方法用于图像配准.首先使用小波变换将源图像处理成不同分辨率图像,再使用改进粒子群算法完成寻优过程,然后将在上层寻优获得的参数作为下层粒子种群中的初始参数,逐层配准,最后实现源图像的精确配准.通过实验发现,本文方法能极大程度降低陷入局部收敛的可能性,健壮性更强,配准结果精确程度可以达到亚像素级.The medical image registration method based on particle swarm optimization algorithm and use mutual information as the measure function has the advantages of high accuracy and strong robustness,but it also has some shortcomings such as local convergence and so on. In order to solve these problems,the self-regulating strategy is introduced into the particle swarm algorithm and the multi resolution method is used for image registration. First,using wavelet transform to stratify the source image. Then using improved particle swarm optimization algorithm for image registration,and take the upper layer registration optimization parameter as an initial parameter of a particle swarm at the next layer. It refined layer by layer,finally achieved accurate image registration. The results show that the proposed method can effectively avoid the local convergence problem,the robustness is stronger,and the registration results can reach the sub pixel level.

关 键 词：图像配准 改进粒子群算法 互信息 多分辨率 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]