Intensity index based histogram equalization technique for retinal image enhancement and classification of hard exudates using supervised learning

Abstract

An efficient, patient friendly method to detect retinal exudates based on binary operation is presented in this study. A novel Histogram equalization technique centered on intensity index is used for fundus image enhancement. Following the elimination of optic disc from the fundus image, morphological operation is performed to detect the exudate pixels. Finally, classification of hard exudates using a trained Support Vector Machine (SVM) classifier is implemented and evaluated using five different performance parameters. The results are assuring and recommends that the proposed method can be utilized as an analytic aid to ophthalmologist for early detection of retinopathy symptoms.