Detecting optic disk based on AdaBoost and active geometric shape model

Detecting the optic disk (OD) is very important in the fundus image analysis. In this paper, we propose a new OD detection algorithm consisting of four main steps: first, obtaining the sub-image which includes the OD from the fundus image based on the saliency map; second, generating the super-pixel from the sub-image with a simple linear iterative clustering (SLIC) algorithm; third, classifying the super-pixel into OD or non-OD based on the AdaBoost algorithm; fourth, fitting the detected OD region with a circle based on the active geometric shape model. The proposed method has been evaluated on the Digital Retinal Images for Optic Nerve Segmentation (DRIONS) database. Experimental results show that our algorithm is very competitive with the state-of-the-art method.