Medical Image Registration Method Based On Characteristic And Gray

Medical imaging has been an important part of modern medical sciences. Medical imaging includes two separate parts: medical imaging system and medical imaging processing. Imaging system lays stress on the imaging acquiring process. Processing is focused on iamge post-processing in order to improve image quality, exert image features or fuse different model of images. With the development of imaging technology, the modality medical images have been used widely in clinical diagnoses and surgical therapies. Integrating those images is helpful to improve the accuracy of clinical diagnoses surgical therapies. Image registration is the key part of integration; it is a geometrical transformation and can rectify two according to the corresponding features.The methods of image registration can be classified into two categories: the feature-based matching approach and the gray-based matching approach. In this dissertation, the characteristics of these two common methods are dissertated and improved algorithms are put forward.The feature-based approach translates the analysis of the whole image into its features, as s result, it speeds up the image process. The exactness of feature detection and orientation has influence on the matching process. In this dissertation, the current feature detection methods are discussed, the approach based on point feature is emphasized.The gray-based approach uses all available gray information of the whole image, so it can improve the precision and robust performance of image registration. Aiming at the shortcomings of mutual information approach, an improved method based on joint histogram is put forward in this dissertation, a novel similarity measure is proposed based on the analysis of the point set distribution of joint histogram. Experiments show that the improved method has better anti-noise performance.Aiming at the shortcomings of the feature-based approach and the gray-based approach, an image registration method based on feature and gray is put forward in this dissertation. Firstly, an improved Harris corner detection algorithm is put forward, by introducing the feature vector, the search of geometric transform parameters is converted into the search of feature points. Then by searching in the neighborhood of initial geometric transform parameters, accurate geometric transform parameters are obtained. The experiments show that compared with traditional methods, the improved method has higher registration precision and stronger applicability and it overcomes the shortcoming of falling into the local extreme value which leads to the error matching well.