Abstract: Retinal image analysis is gaining popularity due to automated disease detection. More and more opthalmists are performing computer aided scanning of the eye. In such an environment, sophisticated feature extraction of retina can lead to better diagnosis of the eye diseases. Retinal image analysis is challenging because different parts of retina like cup, disk and vessels are independent entities and yet affects the detection of one another. Past works of retinal feature analysis is focused towards extraction of either disk area of the retina or the vessels. The study of retina abnormality and its detection is also studied under separate algorithm. Firstly, morphological technique is used to extract the disk ROI. We then locate the disk area by calculating the vessels and determining the bend. An iterative process is adopted for detecting the Cup area within the disk that follows the energy flow towards the bend. Green color channel analysis method is used for vessels extraction. We also compare canny edge based boundary tracing for Disk detection with that of Circularity based boundary tracing technique. Therefore in this paper we develop a framework for automated retinal feature analysis which can detect the retina features like cup, disk, vessels and finding of CDR ratio.

Keywords: Optic cup to disk ratio, Vessel bend extraction, canny edge, Circle based tracking.