| The optic nerve head (ONH) is one of the main features of a fundus image of the retina. Detection of the ONH is a key preprocessing step in algorithms designed for the automatic extraction of the anatomical structures and computer-aided diagnosis of the retina. Digital image processing and pattern analysis techniques to locate automatically the ONH in fundus images of the retina are proposed in the present thesis. The methods are based on edge detection, the Hough transform, Gabor filters, and phase portraits. The center and boundary of the ONH were marked independently by an ophthalmologist for comparison. Free-response receiver operating characteristics analysis as well as measures of distance and overlap were used to evaluate the results. With 40 images from the DRIVE dataset, successful detection rates of 90% and 100% were obtained with the methods based on the Hough transform and phase portraits, respectively. |
