Research Of Fundus Photographic And Angiography Vascular Segmentation And Arteriovenous Classification Methods

With the rapid development of digital image processing,medical image processing has attracted more and more attention.Studies have shown that many systemic diseases in human body are closely related to the abnormalities of fundus blood vessels.However,manual analysis and detection by medical workers are not only time-consuming and labor-intensive,but also difficult to obtain more accurate results.Therefore,it is of great practical significance to achieve automatic segmentation of fundus image vessels and automatic classification of arteries and veins.In this paper,the segmentation of color retinal vessels and the classification of arteries and veins and the segmentation of blood vessels in fundus angiographic images are realized respectively.Fundus camera images of the blood vessels segmentation and classification of arteriovenous part first green channel selection,remove the fundus camera image noise,image enhancement,such as pre-processing,and then USES a filter response based on the combination of conversion and rod-shaped choose combination method to test the blood vessels to dry,using the combination of the asymmetric filter to detect peripheral blood vessels,so as to realize the blood vessels of fundus camera image segmentation,the output as input of subsequent classification.The method using the combination of the response of the combined transform filter and the bar selection has higher accuracy and less time cost than some other methods based on unsupervised learning.Finally,an automatic method of detecting and classifying the arteries and veins of retinal fundus image based on context-related features is used to classify the arteries and veins of color retinal vessels.This paper adopts two methods to realize the fundus angiography image segmentation,adaptive threshold segmentation,the first is a virtual edge tracing method is used to detect the edge of the blood vessels of discrete points,then the image is divided into several sub image,the sub image is consistent,the size of the subsidiary image threshold segmentation,so as to realize the whole image segmentation.The second is the BP neural network,using BP neural network to segment the blood vessel of fundus angiographic image can be divided into the following steps: extracting image information,image transformation,feature extraction,data normalization,optimization processing(using neural network classifier),classification decision and final output.Compared with other traditional segmentation methods,BP neural network is used to segment the blood vessel image more clearly and less noise.Finally,the experiment evaluation and test were carried out on the DRIVE database,which is one of the three internationally recognized fundus image databases.The experimental results showed that the method used in this paper had a good effect on vascular segmentation and arteriovenous classification.