An optic disc localization method based on optic disc appearance characteristics and blood vessel structure

Detection of related tissues in the retinal fundus image of the human eye can diagnose the relevant eye diseases at an early stage. Fundus image processing and analysis is an important application of computer science in the medical field. Through the digital image processing method, automatic detection of the target related to the diagnosis of fundus diseases in the fundus image and screening of fundus diseases can reduce the workload of experts greatly, and is of great significance for the promotion of large-scale fundus screening. This paper explores image preprocessing as well as automatic localization methods suitable for automatic detection of optic disc with high accuracy and sensitivity, which lays a technical foundation for the promotion of fundus screening and large-scale fundus screening. On this basis, this paper studies the method of optic disc localization combined with the appearance characteristics of optic disc and retinal vascular structure, and improve it to test the effect of improved algorithm. Behind, the method of this research is validated and verified by using the open fundus image database. The effect of our method locating the optic disc and the causes of the optic disc localization error are analyzed.