科研之友
微信
新浪微博
Facebook
分享链接
摘要
Medical image stitching plays a critical role in medical diagnosis and treatment, where medical image information of the same or different patterns is merged together for the clinical judgment. The key technique of medical image stitching is to register the sequence of overlaid regions by intelligent approach and to transform them into the unified space to construct a new panoramic image. In this paper, a novel parallel SIFT feature detective algorithm is imported to generate the initial SIFT feature points quickly. Then the transformation fitting optimization problem with a model of point pattern matching is solved by the optimization techniques, where a varietal particle swarm optimization is utilized to obtain the appropriate matched pairs with the optimal transform parameters estimation among the points in the cylindrical space by Levenberg-Marguardt method. The experimental results on both smear and retinal image dataset show that the proposed method is not only effective to both simple and complex medical images, but also shows more accuracy and high efficiency compared to the competitive method.
