序列图像配准方法地研究.pdf

准。 关键词：图像配准、序列图像、角点、互信息、聚类 II Abstract During recent years, along with the rapid development of computer technology, different new technology, new theory and new algorism of computer technology are emerging continuously, and as the basis of image identification, calculation and processing, the image registration technology is also developing rapidly. To register two images, it is necessary to define a registration measure function and find a spatial transformation relation, which maximize the similarity (or minimize the difference) between two images after spatial transformation, i.e. to obtain spatial geometrical consistency between the two images. Though the research on image registration technology has lasted many years, many research results has been achieved, as the difference between imaging mechanisms and conditions among different image instruments, which results in the diversity of input data source of image registration, different applications have different requirements on image registration, which complicate the registration, and sometimes registration results are poor, especially regarding the processing of sequence image registration composed by a large number of images. Therefore, the research on image registration method for sequence image is of great significance. This paper comprehensively and systematically studies the basic theory and mathematics foundation of original research on registration, and brings forward a segmented image registration method combining features and mutual information, on the basis of implementing various experiments on typical operator of such contents used in the image registration process as geometric trans