Diabetic retinopathy is a type of eye disease characterized by retinal damage brought on by diabetes, and is the leading cause of blindness in people aged between 20 to 64 years. Image processing techniques are used to detect and classify retinopathy images effectively. This paper is based on a proposed segmentation algorithm titled “Shrinking Edge-Mark,” which displays two classes of labelled regions, including veins and other regions in the retinal image. From the extracted veins, parameters such as the width and length of each vein are calculated, along with the cotton wool region size and contour of the image. In all, these four parameters are used to detect the stages of diabetic retinopathy. The width and length are calculated from the segmented area by using a basic formula. The area of the cotton wool region is obtained by subtracting the value of the area of the vein region. Finally, a Recursive Support Vector Machine (RSVM) is used to classify the images and accurate detection is successfully determined, notwithstanding the normal or abnormal condition of the retina. MATLAB, a high-performance language for technical computing, is used to implement the concept.