OPTIC DISC DETECTION IN RETINAL IMAGES USING AN EVOLUTION STRATEGY (μ+λ) OPTIC DISC DETECTION IN RETINAL IMAGES USING AN EVOLUTION STRATEGY (μ+λ)

Resumen

In this paper we present an optic disc (OD) detection approach based on evolution strategy (ES). The approach has two main steps: coarse detection and contour edge refinement. Coarse detection estimates an approximate position using an ES in which each individual has a fitness function based on the amount of bright pixels and the number of vasculature structure edge pixels inside a circle. The contour edge refinement uses a geometric approach to circle deformation in order to adjust the circle’s edge to the OD edges. To do so, the pixel with the greatest intensity value variation along a normal line is considered. The proposed approach was evaluated using the STARED and DIAREDB public repositories to process images of normal patients and those with retinal variations. The results show that the proposed method identifies the optic disc position in retinal images with an accuracy of nearly 96%.