Study On The Detection Method Of Macular Fovea In Fundus Images

Central fovea of macula locates on the midpoint of eye bottom and is the central area of opticianry.It is consisted with cone cell and has the optic cells with the highest density,which makes the optesthesia exquisite.Diabetic retinopathy and macular edema mainly appears around the central fovea of macula area.Fundus color photography is one of the major methods to diagnose ophthalmic disease.Therefore,precise location of macula fovea is significant in the retinal image analysis of arteriosclerosis,hypertension and diabetes.The localization method using wavelet transform and gray contour for macula fovea is proposed.Firstly,the modulus minimums are extracted as the candidates ofmacula fovea center by the quadratic spline dyadic wavelet transform without down-sample.Secondly,the gray contour is structured based on the Gaussian-like grayscale distribution characteristic of macular to detect the fovea.This method can locate the macula fovea accurately and effectively.The test result on 40 fundus images of DRIVE database demonstrates that the method achieves an accuracy of 100%.The hospital graphic databases shows that the accuracy rates of diabetes are 95.12%and 91.43%in phase 1 and phase 2 respectively.