青光眼和糖尿病视网膜病变可以通过视盘来反映情况, 自动快速的定位和分割视盘无疑给眼科医生带来了便利。提出了基于多圆分割视盘的方法, 利用视盘具有高亮度和垂直血管发布的特点定位到视盘中心, 对形态学方法移除血管后的视盘区域进行Hough圆变换, 对检测到的多个圆进行半径调整, 通过计算多圆的最小凸包从而分割出视盘。在DRIONS-DB数据库上, 该算法准确定位到视盘中心, 并且分割的平均覆盖率、Dice系数、精确度、灵敏度、特异性分别达到87.97%, 93.33%, 99.61%, 94.74%,99.75%, 平均处理时间仅为1.1s, 表现出较高的准确性和快速性, 适合自动化的视盘分割处理。

Glaucoma and diabetic retinopathy can reflect the situation through the optic disc, automatic rapid positioning and splitting the optic disc undoubtedly brought ophthalmologists convenience. A method based on multi-circle segmentation is proposed. The center of the optic disc is located by using the features of high brightness and vertical blood vessels in the optic disc. Hough circle transformation is performed on the disc area after the vessel is removed by the morphological method. A circle for radius adjustment, by calculating the minimum circle of multi-convex convex so as to divide the disc. On the DRIONS-DB, the algorithm accurately locates the center of the optic disc, and the average coverage,Dice coefficient,accuracy,sensitivity and specificity of the segmentation reach 87.97%, 93.33%, 99.61%, 94.74% and 99.75% Time is only 1.1s, showing a high degree of accuracy and speed, suitable for automated video disk segmentation.