Research On The Recognition Of Glaucoma Image Based On Image Segmentation

Glaucoma is one of the most common disease that cause blindness in modern society, which is characterized by elevated IOP,atrophy of the optic nerve and defect of visual field. Glaucoma makes a serious threat to our health, even to the development of our society. It can make very serious harmful consequences. Because of the secretiveness and progressive of the glaucoma,once detecting decline of eyesight,it's a late stage, Then the defect of vision field is serious and cannot be recovered.In view of the above, earlier discovery and earlier treatment is very important.In the rapid development of information today, the identification of glaucoma is not perfect, and the recognition rate is not high, and even cannot meet the needs of the community.thus, the research of computer recognition is very significant.Firstly,achieving the segmentation of eye-ground image of both disc and cup based on Markov Random Field (MRF) theory. Then,we can delineate the boundary of both disc and cup by using a minimum perimeter of polygons (MPP) approximately. Finally, achieving the recognition of eye-ground image of glaucoma by the Euclidean distance identification method and by the correlation coefficient identification method that based on the ISNT rule. Image segmentation based on Markov Random Field is based on statistical theory.Although MRF itself have not the segmentation capabilities and can only provide a segmentation model.It concerns the correlation between the local pixels,uses the priori knowledge effectively,has fever model parameters and is easy to be combined with other methods etc.so this method has been extensive researched and application in the field of image segmentation.The main task of this thesis can be summarized as the following two parts:1?the segmentation of eye-ground image of both disc and cup and delineate the boundary of them:This paper introduces the relative theory of MRF, then, makes a MRF-MAP model by MRF theory, at the same time, combining k-mean segmentation algorithm, EM, iterated conditional mode algorithm (ICM) and other algorithms that achieve the segmentation of eye-ground image of cup and disc. Finally, achieving the boundary of both disc and cup by using a minimum perimeter of polygons (MPP) approximately.2?The recognition of glaucoma:Achieving the recognition of glaucoma image by the Euclidean distance identification method and by the correlation coefficient Identification method that base on ISNT rule.