Automated localization of optic disk in retinal fundus images using cluster region membership and vessel network

摘要

In this paper, a fully automated approach for optic disk localization in the retinal fundus images has been presented, which is completely independent of any specialist intervention. The proposed method tries to localize the optic disk region based on a parameterized membership function defined on the cluster regions and the predicted convergence point of the retinal vasculature. The proposed algorithm has been implemented using partition around medoids as well as k-means clustering technique, and performance of both the clustering approaches have been analyzed. The proposed algorithm has been tested on four publicly available fundus image databases. In localizing the optic disk, our proposed method has demonstrated highest average overlap, mean absolute distance and accuracy for DRIVE database and best average overlap, processing time and accuracy for DIARETDB1 database, respectively, for its partition around medoids implementation and best average sensitivity for its k-means implementation, outperforming most of the reviewed state-of-the-art methodologies available in literature.