自适应光学高分辨率视网膜图像复原

人眼视网膜是一个结构复杂的人体组织,不少眼部和非眼部疾病都与视网膜密切相关,同时能在视网膜的图像中反映出来。人类对人眼视网膜成像的研究也由来已久,如何获取高分辨率的视网膜图像,为视觉生理研究和疾病的早期诊断提供前所未有的有力工具是该领域的一个研究热点。以此实际问题为研究背景,针对不同的图像复原条件及难点,本文分别采用LR方法非盲复原、超变分正则化盲复原法、变分Bayes估计算法等方法,并在此基础上对模型进行相应改进后对图像进行复原。

the retina of the human eye is a complex human tissue,many eye and non-eye diseases are closely related to the retina,and can be reflected in the image of the retina.The research of human eye retina imaging has a long history.How to obtain high resolution retinal images and provide unprecedented powerful tools for visual physiology research and early diagnosis of diseases is a research hotspot in this field.Based on this practical problem,this paper adopts LR method for non-blind restoration,hypervariational regularization blind restoration,variational Bayes estimation algorithm and other methods for different image restoration conditions and difficulties.On this basis,the model is improved and the image is restored.