Automatic Detection of Hard Exudates in Retinal Images Using Haar Wavelet Transform

ABSTRACT Diabetic Retinopathy (DR) is a leading cause of vision loss, caused by the abnormalities in the retina due to insufficient insulin in the body. So that Diabetic patients require regular medical checkup for effective timing of sight saving treatment. A completely automated and robust screening system for the detection of Diabetic Retinopathy can effectively reduces the burden of the specialist and saves cost as well as time. Due to noise and other disturbances that occur during image acquisition, DR may lead to false detection and this is overcome by various image processing techniques. This paper presents an automated method for bright lesions i.e. hard exudates in retinal images. The Haar wavelets transform is used in the system for the hard exudates segmentation followed by k nearest neighbor classification method. We have used four databases of fundus images; among them we obtain sensitivity 37.14%, 21.87%, 12.50%, 25.47% for MISP, DB0, DB1 and STARE database respectively. And the specificity is 0% for MISP and 1% for remained databases.