Retinal Vasculature Segmentation

Abstract

To identify various retinal pathologies like diabetic retinopathy, hypertension, glaucoma, arteriosclerosis fundus vasculature plays important role.  To assist specialists for the determination of diseases it is essential to extract the retinal vasculature. This paper represents a method to segment the retinal vasculature. This paper contains four parts : first part is  preprocessing  which prepare funds images to extract vessels. Secondly an image enhancement  includes bottom-hat transform followed by filtering. In third stage K-means is used  to  acquire binary  segmented image; and lastly wrongly selected and detached group of pixels removed in postprocessing step. Online available databases STARE and DRIVE  are used to test the method.