Fundus images are valuable resources in diagnosis of retinal diseases. This paper proposes a computer-aided method based on various feature extraction techniques and support vector machines (SVM) for detection and classification of diabetic maculopathy (DM). DM, defined as retinopathy within one disc diameter of the centre of the macula, is a major cause of sight loss in diabetes. Here, we bring out a new approach to detect DM based on retinal fundus image features. During the first stage the input image is enhanced and the optic disc is masked to determine the presence of regions of foveal neighborhood. The second stage, deals with various feature extraction technique based on transform, shape and texture features. Extracted features are further categorized as healthy or affected images. Here we go for classification task using the RBF Support Vector Machine (SVM) classification, the techniques have been tested on retinal databases and these are compared with trained phase to categorize Healthy and DM images. This method can detect DM with a level accuracy on par with human retinal specialists
Determination of cup-to-disc ratio of optical nerve head for diagnosis of glaucoma on stereo retinal fundus image pairs
