Abstract:
Diabetic complication causes variation in the retinal blood vessel, channels to Diabetic Retinopathy (DR) and the retinal blood vessel gets distorted. The exudates are concealed; microaneurysms and hemorrhages arise in the retina. These features represent the disease severity. DR eye disease which is harmful causes pressure in eye nerve fiber. It is essential to diagnose it earlier. The proposed approach is used for the detection of severity of DR. The proposed approach starts with preprocessing using hybrid enhancement technique, reconstruction filter segmentation, extraction of Gabor Features (GF) and Histogram of Oriented Gradients Features (HOGF), classification and optimizations. Publicly available databases such as DRIVE, STARE, MESSIDOR and Real time fundus images are taken for analysis for performance comparision.
