一种基于高斯匹配滤波法的视网膜图像血管中轴线快速提取算法

A Robust Method For Fast Detection of Vessel Skeleton in Retinal Images Based on 2 D Gaussian Matched Filters

目的：研制快速、精确的视网膜血管中轴线（ｖｅｓｓｅｌａｘｉａｌｓｋｅｌｅｔｏｎＶＡＳ）提取算法，以定量分析眼底血管形态改变的细微变化。方法：１．图像预处理（去噪、增强）。２．以经二值化及膨胀后的梯度图像为模板，进行基于高斯匹配滤波法的快速血管提取。３．后处理（细化等）。结果：分别采用本文方法及传统方法对视网膜ＶＡＳ进行提取与结果比较。结论：本文方法具有快速、精确。

Purposes:This paper puts forward a robust method of fast detection of vessel skeleton in retinal images for quantitatively analysing subtle morphological changes in retinal vessels.Methods:1 Image preprocessing(De noising,Image enhancement).2 Fast detection of retinal vessels was achieved by performing 2 D Gaussion matched filters only on the mask images (binarized and dilated gradient image).3 Postprocessing(vessel thinning,et al).Results:The performance of the method was compared with those of the traditional methods and found it to be superior.Conclutions:VAS can be fast detected with continuous structure of the vessel,less sensitivity to noise, and high accuracy on conventional computers.