Bright region and vessel density based robust optic disc segmentation

Abstract

The precise segmentation of optic disc is a major task in developing automated systems for diagnosis of retinal diseases using colored fundus images. The noise, illumination effects, blood vessels and lesions like exudates make it really challenging to build an efficient OD segmentation method using fundus images. This paper presents a novel method for accurate segmentation of optic disc in the presence of noise and other lesions. The proposed methodology comprises of four modules i.e preprocessing, vessel segmentation, bright region extraction and detection of optic disc. The proposed system is tested and validated on publicly available fundus image databases and images gathered locally for real images. The experimental results show the validation of the proposed system.