The Key Technologies Of Fundus Camera Image Processing And Analysis

Fundus has been a clinical close attention to areas that are often associated with systemic retinopathy is closely related to other diseases. Number of important systemic diseases may be caused by a specific reaction or retinal complications in clinical practice can often be some special signs from the fundus of the other disciplines to help make the correct diagnosis and prognosis estimation. Fundus is also the"barometer", reflecting diabetes and high blood pressure condition change. By observing the various symptoms of retinal arteries, such as arterio-venous diameter, the wall changes ,the phenomenon of cross-oppression and extent of retinal hemorrhage, exudation and the nature of the presence or absence, to a certain extent, understand the situation of systemic vascular.As the fundus image is more complicated, some of the traditional image processing methods can not satisfactorily resolve fundus image enhancement, image segmentation and lesion recognition problems and so on. This needs to be developed for a variety of fundus image processing algorithms in order to truly achieve fundus image analysis and research. Based on the access to a lot of information and advice on the medical experts, and combined with research group is developing a certain type of digital fundus camera image processing system , this paper study the key technologies in-depth of fundus image processing and analysis:(1)Fundus image enhancement. The process of fundus image capture,transmission and storage , due to objective conditions of the restrictions will result in lower quality images.This paper selecte the grayscale fundus images to enhance by gray-scale transformation, and propose the fundus image enhancement methods based on rough set and wavelet unsharp masking. The enhanced fundus images are more conducive to a doctor's interpretation, and follow-up image processing provides a good guarantee.(2)Fundus lesion identification. For each period of diabetic retinal disease lesions, based on the threshold method, the paper apply the Otsu threshold method and iteration method to segmentate fundus grayscale images. The identified Fundus lesions will help identify the doctor's diagnosis, thus facilitating the diabetes diagnosis of doctors.Take C++ Builder and Access as platforms, it has developed fundus image processing and analysising system. The paper has a positive meaning for software systems of fundus camera. Meanwhile it provides reference for subsequent image processing and analysis of computer-aided diagnosis.