中文摘要：

糖尿病性视网膜病变进行早期筛查可以减少疾病的发展并且阻止随后的视力损害。微血管瘤是糖尿病性视网膜病变的早期临床症状，可以通过微血管瘤检测对糖尿病性视网膜病变进行早期筛查。针对眼底图像中视网膜血管、视盘、渗出物以及微血管瘤之间的相互关系，在红色通道和绿色通道加权图上定位出视盘，在绿色通道上采用基于简单统计的自适应双阈值Canny算子进行边缘检测，并进行封闭区域的填充。设定阈值消除大面积对象并移除视网膜血管、视盘和渗出物得到微血管瘤的候选区域，最后根据形状特征和颜色特征从候选区域中得到真正的视网膜微血管瘤。实验结果表明，该算法能够有效提取视网膜眼底图像中的微血管瘤，敏感性和阳性预测值分别达到92%和86%，优于现有一些典型的微血管瘤检测方法，能够精确地检测出微血管瘤，可用在糖尿病性视网膜病变早期筛查中。

英文摘要：

The early screening of the diabetic retinopathy can restrain the development of disease and prevent the subsequent vision impairment.The microaneurysm is the earliest clinical sign of the diabetic retinopathy,and its detection can perform the early screening for the diabetic retinopathy.Considering the interrelation among the retinal blood vessel,optic disc,exudates and microaneurysm in the eye ground image,the optic disc is located in the weighted images of the red channel and green channel.The adaptive dual?threshold Canny operator based on simple statistics is adopted in the green channel to perform the edge detection,and fill the enclosed region.The threshold is set to eliminate the large area objects,and remove the retinal blood vessel,optic disc and exudates to acquire the candidate area of the microaneurysm,in which the real retina microaneurysm is obtained according to the shape feature and color feature.The experimental results show that the method can extract the microaneurysm in the retina eye ground image,the sensitivity and positive predictive values can reach up to92%and86%re?spectively,the method is superior to some typical microaneurysm detection methods,can detect the microaneurysm accurately,and is useful for the early screening of the diabetic retinopathy.