Retinal Vessel Segmentation Based On Level Set Method With CGLI Model

As a non-intrusive method, the retina imaging provides us a better way for the diagnosis of ophthalmologic diseases. Extracting the vessel profile automatically from the retina image is an important step in analyzing retina images for computer aided diagnosing.We propose a hybrid active contour model to segment the fundus image automatically. We combine the signed pressure force function introduced by the GCV model with the local intensity property introduce by the LSI model to overcome the difficulty of the low contrast in the segmentation process. Firstly, a global and the local information are used to redesign the signed pressure force function (SPF). Secondly, the local and the global information are also used to ensure the robust of our method?Thus our method is called CGLI(Combined Local and the Global Information)Experimental results on the public a publicly available DRIVE database (se=0.7058, sp= 0.9699, acc=0.9360) show that our method is effective in retina image vessel segmentation and is robust to some of pathology images compared with the traditional level set methods.