A fast and efficient approach for 10calizing and segmenting也已optic disc in fluorescein retina1 images by means of mathematica1 m。中ho1ogyand也已gradient vector flow (GVF) snake mode1 is proposed. The10calization a1gorithm utilizes the similarity in gray intensity between b100d vesse1s and也已optic disc. In也已approach，也已retina1 images are first enhanced using two top-hat operators， where the optic disc is 10calized using Ot間's thresho1d and image sub仕action. Aft已r 10calization， pixe1-1eve1 preprocessing is performed to remove b100d vesse1且， whe間the optic disc boundary is detected using the GVF snake model. The proposed method is eva1uated using a database of 60 fluorescein images. The accuracy rate of disc 10calization is 96.7%立起boundary detection method has average sensitivity va1ues of 95.1 % for images with a defined optic disc and 90.4% for images without a defined optic disc.