基于混合优化算法的医学图像配准技术  

Optimization Medical Image Registration Based on Mixed Optimization Algorithm

作　　者：王雪[1] 王秋爽[2] 

机构地区：[1]吉林农业科技学院教育技术与信息中心,吉林吉林132101 [2]吉林大学计算机科学与技术学院,长春130012

出　　处：《吉林大学学报（信息科学版）》2016年第5期703-708,共6页Journal of Jilin University（Information Science Edition）

基　　金：吉林省自然科学基金资助项目(20150101055JC)

摘　　要：在图像配准的优化算法中,为避免使算法陷入局部最优的问题。因此,提出基于最大互信息和混合优化算法的医学图像配准算法,利用模拟退火算法思想改进粒子群优化算法,提高了全局寻优的能力,能更好地跳出局部最优。由实验结果可知,该方法不仅具有较好的图像配准精度,对椒盐噪声和高斯噪声也有较好的鲁棒性。Medical image registration technique is floating medical image and the reality of physical space match point, which helps doctors to quickly find the lesion area. In the image registration optimization algorithm, if the optimal value is always the same, it will be obtained locally optimal solution. Therefore, we propose optimization algorithm based on mutual information and mixed medical image registration algorithm, and improves the ability of global optimization based on this technology. Experimental results show that it has better image registration accuracy, and has robustness on salt and pepper noise and Gaussian noise.

关 键 词：医学图像配准 互信息 粒子群优化算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]