ANALYSIS OF SUPERPIXEL AND PRINCIPAL COMPONENTS FOR OPTIC DISC SEGMENTATION

In this project an automatic segmentation tool for the detection of optic disc is used to assist clinician to prevent visual loss due to
DIABETIC retinopathy, hypertension, glaucoma, and macular degeneration. A Superpixel generation and Principle Component
Analysis (PCA) based algorithm is proposed for Optic Disc (OD) segmentation. It makes use of different operations such as
generalized distance function (GDF), a variant of the watershed transformation, the stochastic watershed, and geodesic
transformations. The Optic Disc (OD) segmentation is done in three steps. In the first step, RGB fundus image is acquired from patient
data base and the image is pre-processed by superpixel generation, to divide the image into superpixels and by PCA, where the gray
level image is obtained. In second step, it employs the gray-image centroid and Stochastic Watershed transformation is used. In the
third step, Circular Approximation is done in Post processing process and the Optic Disc-contour has been estimated.


!PDF