Effective contrast enhancement techniques for Fundus images

Abstract

Type-I and Type-II diabetes is an ever increasing condition booming nowadays in this era. Diabetic Retinopathy (DR) is an eye condition which is targeted on such patients. The characteristic of this DR is basically aimed at the minute blood vessels present in the eye. A feature known as “Exudates” starts to appear while examination of the eye. The cause of such features in the fundus of the eye is actually the presence of fluid that would eventually ooze from the vessels. Neglecting this situation would result in blindness too. A detailed analysis is necessary for us to improve the patient’s health. Examining the fundus of the eye and analysing it accurately would help in avoiding the severity of the disease. Image processing technique allows one to capture an RGB model of our fundus wherein all the features of the eye, that might include blood vessels, exudates, and haemorrhage if any. With the obtained image, the major job is to now analyse the data thoroughly and provide an appropriate treatment. In order to distinguish between the exudates and the other features, we first choose an effective colour model to observe a better contrast of the exudates. Further, we perform a unique segmentation technique wherein we automatically detect the adversity of the exudates population in the fundus image of the eye by which we would be a guide for suggestive treatments.