【摘要】 提出一种基于互信息的视网膜眼底图像配准方法 ,并采用遗传算法求解待配准图像的变换参数。与传统方法相比 ,该方法具有配准精度高、可靠性好、不需要进行图像的预分割和特征提取等特点 ,配准精度可以达到亚像素级。更多还原

【Abstract】 A new registration method based on mutual information has been proposed for retinal fundus images, using genetic algorithms to solve transformation parameters. Compared with the traditional methods, the mehtod has some advantages such as higer precision and reliability as well as no need of image presegmentation and feature extraction. The implementation of this method is presented in detail in this paper, and its accuracy is sub pixel level.更多还原