Retinal fundus photographs has always remained the gold standard for evaluating the changes in retina. Here, a novel method for automatic glaucoma detection from digital retinal fundus images is proposed. The methodology makes use of optic disc and cup segmentation. Optic disc is segmented using morphological operations and hybrid level-set methodology. Optic cup is segmented by first detecting blood vessels using SVM classifier and then the bending points on the circum linear vessels.
Parameters such as vertical cup-to-disc ratio (CDR), cup-to-disc area ratio are calculated and used for glaucoma detection. A CDR value greater than 0.5 and cup-to-disc area ratio greater than 0.3 indicates the presence of glaucoma. The proposed method is found to produce a mean error as low as 0.021 (CDR) when compared with expert observation.