摘要
基于互信息的三维医学图像配准是配准的一大热点.本文提出的配准方法主要是对互信息测度的公式进行简化,修正了Powell算法的方向替换方法,保持了原搜索方向,提高了配准速度.合理的设置收敛阈值,并用多分辨率策略进行了图像配准测试.实验表明,改进后的互信息测度可以达到亚像素精度,而且在配准速度上有了明显提高.
Medical image registration based on Mutual Information(MI) is a hot topic in registration research. The MI's formula is modified to reduce computing cost, and the replace of Powell's, search direction is also rectified to reserve the original direction, Multi-resolution method is used in our registration test. The results show that the modified MI measure can reach sub-voxel precision, and the registration time is reduced remarkably.
出处
《山东大学学报(工学版)》
CAS
2006年第4期33-36,113,共5页
Journal of Shandong University(Engineering Science)
基金
科学技术发展计划项目(2005GG4201010)