Retinal vessel extraction by combining radial symmetry transform and iterated graph cuts

In this paper, we propose a new method for the extraction of blood vessels in retinal images. This approach starts with a Hessian-based multiscale filtering method to enhance blood vessels in gray retinal images. Subsequently, a new radial symmetry transformation, which is based on line kernels, is proposed to improve the detection of vessel structures and restrain the response of nonvessel structures. Finally, an iterated segmentation algorithm is used to extract retinal vessels. The proposed approach has been tested on the two publicly available databases, DRIVE and STARE. The experimental results show the feasibility of the proposed method.