摘要
针对目前基于互信息图像配准方法中存在的不足,提出结合图像二维信息(如轮廓和边缘)与互信息的图像配准算法。它首先利用小波多尺度积提取两幅图像的特征点及其角度信息,再根据得到的特征点和角度信息,定义了特征点对互信息匹配准则,得到相应的匹配点对。最后进行了仿真实验,并将结果与由相关度、对齐度准则得到的结果进行比较,所提出的算法匹配误差最小。结果表明该算法具有匹配精确、鲁棒性好和效率高等优点。
To address the disadvantages of mutual information-based image registration, a new method for image registration is proposed that combines image edges with mutual information to form a new matching criterion. In this method, feature points as well as their orientation are extracted using wavelet multi-scale products and the mutual information mapping criteria for the feature points is defined to obtain the matching points. The experimental results show that this method yields the minimum matching error compared with that of the correlation and alignment criteria, which demonstrates that the algorithm is accurate, robust and efficient for image registration.
出处
《重庆大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2009年第6期697-700,704,共5页
Journal of Chongqing University
基金
国家青年自然科学基金资助项目(60604007)
重庆市自然科学基金资助项目(CSTC2009AC2057)
关键词
图像配准
特征点
互信息
多尺度积
image registration
feature point
mutual information
multi-scale product
