MICROANEURYSM AND EXUDATE DETECTION FROM COLOR FUNDUS IMAGES USING VARIOUS IMAGE PROCESSING TECHNIQUES

Abstract

This paper proposes technique for improving the contrast and sharpness and reducing noise of fundus images and methods for detection of Non-proliferative Diabetic Retinopathy symptoms from the preprocessed fundus image. Image Pre processing is done using Dualistic Sub-Image Histogram Equalisation method. After pre-processing the optic disk and exudates in retinal images are segmented out using contour and morphological techniques. Once the contour of a given pattern is extracted, its different characteristics will be examined and used as features which will later on be used in pattern classification. Microaneurysms are segmented out using various morphological techniques from the given preprocessed retinal image. On analyzing the results of the above algorithms, it concludes the presence or absence of DR symptoms in the given retinal image.