| In recent years,incidence of diabetic mellitus is increasing year after year with rapid development of society and changes of people's life style and people pay more attention to diabetic mellitus than before.Diabetic Retinopathy(DR),as one of the most common micro-vascular complications,can lead to decreased vision and lead to blindness in severe cases.Hemorrhage(HA)is one of obvious early symptoms of DR,so accurate detection of HA has important significance in building automatic screening system of DR.The disadvantages of mainly existed methods are false detection and missed detection of HA because the methods only take into account Gaussian characteristics of HA,and not grasp essential differences between HAs and non-HAs,such as blood vessels,background mutation and noise.In order to solve the problem of false detection and missed detection of HA,a method based on two-dimensional Gaussian fitting and human visual characteristics is proposed in this thesis.Firstly,brightness correction and contrast limited adaptive histogram equalization are used to preprocess original color fundus image.Secondly,candidate hemorrhages are extracted based on background estimation and watershed segmentation.Thirdly,2D Gaussian fitting and human visual characteristics are used to extract visual features of candidate hemorrhages.Finally,hemorrhages are obtained from candidate hemorrhages based on visual features.The proposed method is evaluated on 219 fundus images of DIARETDB database.Experimental results show that overall average sensitivity,specificity and accuracy for hemorrhage in image level are 100%,80%and 90.87%respectively,and overall average sensitivity and positive predictive value for hemorrhage in lesion level are 98.52%and 92.30%respectively.Performance values of the proposed method are higher than existed representative methods.The results show that the proposed method can realize automatic detection of hemorrhages in fundus image with high accuracy. |
PDF Full Text Request