眼底图像畸变的校正及多幅图像的特征点拼接算法

DISTORTION CALIBRATION AND MERGENCE ALGORITHM OF MULTIPLE OCULAR FUNDUS IMAGES

对于眼底图像 ,由于不在正视点取像造成的几何畸变和由于荧光素从血管壁渗漏等造成的灰度畸变 ,几乎是不可避免的。为临床诊断提供丰富信息的需要出发 ,进行多幅眼底图像的拼接是必要的。本文为此提出了几个新的处理方法 ,如基于大窗口的滤波灰度校正方法、基于透视投影原理的几何校正方法与基于小波子空间上的拼接算法 ,均具有效果好和耗时少的优点。

For ocular fundus images (OFIs), geometric distortion due to taking pictures at oblique visuai points and gray level distortion due to leakage of fluorescein and so on are almost unavoidable. To provide better information for clinical use, it is very necessary to match many OFIs as a big one. In this paper, several new methods, such as gray level calibration based on filtering technique with large window, geometric calibration based on principle of perspective projection, and registration scheme based on wavelet multi scale decomposition of images, are proposed, all of which have advantages of better effect and less computational time.