Analysis of Retinal Vasculature by Watershed Segmentation and Histogram Analysis

Retinal image analysis is becoming eminent as a nonintrusive diagnosis method in modern ophthalmology. This paper is mainly focused on the early diagnosis of diabetic retinopathy by analysing and detecting of vascular structures in retinal images. When small vessels in the retina have high level of glucose, it produces blur vision which eventually leads to blindness. Usually retinal images are taken from DRIVE dataset. The small vessels which are abnormal are not visible by naked eye are segmented accurately by watershed segmentation. Through watershed segmentation we enhance the blood vessel and suppress the background information. The segmented abnormal nerve image is compared with the healthy and normal nerve image through histogram equalization. Experimental results achieved from the proposed method effectively used to reduce the time for the ophthalmologist to detect disease and give accurate treatment.