基于粒子群优化与梯度下降法的多模图像配准  

Multimodal image registration based on PSO and gradient descent method

作　　者：赵于前[1] 刘彬旭[1] 李桂源[2] 

机构地区：[1]中南大学地球科学与信息物理学院,湖南长沙410083 [2]中南大学肿瘤研究所,湖南长沙410078

出　　处：《光电子．激光》2011年第6期940-944,共5页Journal of Optoelectronics·Laser

基　　金：中国博士后科学基金特别资助项目(200902482);湖南省科技计划资助项目(2009FJ3015);湖南省自然科学基金资助项目(09JJ3119)

摘　　要：针对多模图像配准单纯使用粒子群优化(PSO)算法导致搜索速度慢这一现象,提出了结合PSO和梯度下降法的分步式搜索算法。以CT与MR图像配准为例,计算了使用互信息作为相似性测量时的配准函数,进而分析了图像配准算法中可能遇到的局部极值问题。根据配准函数的特点,提出在局部极值较少时使用梯度下降法代替PSO算法搜索以减少运算量。实验结果表明,对初始位置偏差较大的CT与MR图像,本文算法能够成功地进行配准并能达到理想的精度,与单纯使用PSO算法相比,算法的效率有一定的提高。To improve the efficiency of particle swarm optimization（PSO） used in the multimodal image registration,a new method combining PSO and gradient descent method is proposed.Taking the registration of CT and MR images as example,the method calculates the registration function which uses the mutual information as the similarity measurement,and analyzes the local minima problem during searching for the minimum point.According to the characteristics of registration function,the method uses gradient descent method instead of PSO to reduce the computation when the local minima problem is not serious.The experimental results show that the proposed algorithm is efficient and accurate even if the initial position error between CT image and MR image is large.The algorithm is also more efficient than that only using PSO.

关 键 词：图像配准 粒子群优化(PSO) 梯度下降法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]