一种基于角点特征的医学图像配准方法  

A medical image registration method based on corner feature

作　　者：吕晓琪[1] 黄显武[1] 张宝华[1] 

机构地区：[1]内蒙古科技大学信息工程学院,内蒙古包头014010

出　　处：《内蒙古科技大学学报》2010年第1期49-52,共4页Journal of Inner Mongolia University of Science and Technology

基　　金：国家自然科学基金资助项目(60761002)

摘　　要：医学图像配准技术是医学图像融合及影像三维重建的基础.提出了一种基于SUSAN算子的特征点的医学图像配准方法.该方法利用SUSAN算子提取两幅图像的角点,通过粗匹配和细匹配两个步骤建立特征点集之间的对应关系,得到匹配特征点对.然后再根据这些特征点使用反向映射的仿射变换实现图像的配准.实验表明,此方法能有效地实现医学图像的配准,并取得良好的精度.Medical image registration technology is the foundation of the medical image fllsion and it is also the element of the imagebased three-dimensional reconstruction. A medical image registration method was presented based on the feature points calculated by SUSAN operator. First, the corner points of two images were extracted with the SUSAN operator; then, the two feature points sets were matched and the matched feature point pairs were obtained through coarse matching and fine matching;Finally,the reverse mapping of the affine transformation were utilized to achieve image registration. The experimental results show that the medical image registration and good accuracy can be effectively achieved by using this method.

关 键 词：特征提取 SUSAN算法 角点匹配 距离约束 图像配准 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]