Detection of optic disc and macula and their evaluation in retinal fundus images

This work proposes simple and efficient methods for detection, segmentation, and measurement of the optic disc and localization of the macula. The thesis also proposes methods for calculating the diameter of optic disc. A tool is provided for verification of the location of optic disc and macula to increase reliability of the system. The method proposed here differs from existing methods in that it applies inverse segmentation, automatic extraction of background image and such.Another section of the thesis proposes some methods to detect and segment some kinds of diabetic retinopathy lesions in retinal fundus images. This is extremely important since most of the vision loss in recent years stems from diabetic retinopathy. Additionally, another method proposed in scope of the work is that a simple but very successful biometric recognition system was proposed based on retinal vessel segmentation. This method shows a quite good performance compared to the existing methods.The system is tested by using retinal fundus image data obtained from KTÜ, STARE project?s database and the results are promising. Diabetic retinopathy detection and biometric recognition system proposed here also give very successful results.