基于动态阈值选择和细化算法的角膜细胞轮廓抽取  

Contour Detection Based on Fusion of Adaptive Thresholding and Thinning Algorithms for Corneal Endothelium Images

作　　者：赵立初[1] 施鹏飞[1] 

机构地区：[1]上海交通大学图象处理与模式识别研究所

出　　处：《上海交通大学学报》1998年第9期67-69,共3页Journal of Shanghai Jiaotong University

摘　　要：角膜细胞图象轮廓提取的关键是分隔出细胞与背景，生成二值图象．由于角膜细胞图象的光照不均匀，用单一域值无法兼顾亮区与暗区．提出了一种基于动态域值选择和细化算法的角膜细胞轮廓抽取算法．首先对处理图象进行灰度平均和中值滤波平滑处理，消除细胞内部的不均匀性和保持细胞之间的边界；然后用动态域值选择算法分割细胞和背景；最后使用细化算法抽取细胞的轮廓．此算法应用于角膜细胞图象，处理结果证明了算法的有效性．The crucial procedure of contour detection for corneal endothelium image is to segment cells from background. The singular threshold is uneffective because the image is very different in illumination. In this paper, one corneal contour extraction algorithm based on the fusion of the adaptive thresholding and thinning algorithm is presented. First, the average grey and median filter smoothing algorithms are used to eliminate the heterogeneity and preserve the contours between the cells; then the adaptive thresholding is used to segment the cell regions and the background. Finally, the thinning algorithm is adopted to extract the contour. The effectiveness of the proposed method is validated by applying it to the images.

关 键 词：角膜细胞 图象处理 轮廓抽取 角膜疾病 诊断 

分 类 号：R772.2[医药卫生—眼科] R770.43[医药卫生—临床医学]