Optic Disc Detection in Retinal Images using Tensor Voting and Adaptive Mean-Shift

Abstract

In this paper, we present a new approach to identify the position of an optic disc in retinal fundus images. The proposed algorithm is based on a tensor voting to analyze vessel structures, and the position of the optic disc is identified by mean-shift based mode detection. The core technique consists of three preprocessing stages that enhance local contrasts through illumination equalization, and we extract vessel patterns by tensor voting in equalized images. Lastly, the position of the optic disc is identified by mode detection that is based on mean-shift procedure. Our approach is evaluated against a dataset including healthy retinal images and diseased retinal images. The experimental results show that the proposed method can exactly identify the position of the optic disc from retinal fundus images.