An effective approach to detect hard exudates in color retinal image

Abstract

Detection of hard exudates from fundus images is crucial since hard exudates are considered to be one of the most prevalent earliest signs of retinopathy. To overcome the obstacles in retinal exudates identification, such as: wide variability in color, illumination uneven. An effective approach is proposed. After preprocessing, the histogram thresholding is used to recognize the background and object, and then the Fuzzy C-Means(FCM) technique is applied to assign the pixels remain unclassified in the last stage. The algorithm performance was assessed using a Standard Diabetic Retinopathy Database DIARETDB0. The proposed algorithm obtains a sensitivity of 84.8% and a predictive value of 87.5% using lesion-based criterion, The experimental results show that the proposed approach can detect hard exudates effectively. © 2012 Springer-Verlag GmbH.