Digital Retinal Fundus image is analyzed for the classification and stages of Diabetic Retinopathy (DR).This imparts importance, since many of heart, lung, and kidney related problems could be predicted well in advance, by analyzing the fundus image itself which is an cost effective technique. Also, Later stages of DR causes abnormal changes in human retina and vision loss is occurred in most of the patients. The aim is to extract the abnormal features like Microaneurysms, exudates, and blood vessels, to classify DR at its proliferative stage itself. At the same time, the healthy or normal features like optic disk, blood vessel map is to be removed by suitable enhancement technique (Morphological technique) .The abnormal features and its normal ground features are trained for classification. The classification is done using ANFIS architecture. The detection and classification method , surely provide an additional and promising data and information to ophthalmologists and to the analyst for further treatment. The proposed method will give a promising accuracy when compared with the methods available in the literature.