Retina Vessel Detection Using Fuzzy Ant Colony Algorithm

Abstract

Vessel extraction in retina images is a primary and important step in studying diseases including vasculature changes. In this paper, a fuzzy clustering method based on Ant Colony Algorithm, inspired by food-searching natural behavior of ants, is described. Features of color retina images are extracted by eigenvalues analysis of Hessian matrix and Gabor filter bank. Artificial ants in the image use these features for searching and clustering processes. Experiments and results of proposed algorithm show its good performance in vessel extraction. This algorithm is tested on DRIVE database and its results are compared with other works using the same database. The accuracy of our method is 0.933 versus 0.947 for a second observer.