基于图像分割和盲反卷积的透射电镜图像漂移校正方法  

The method for TEM image drift correction based on image segmentation and blind deconvolution

作　　者：吕龙飞 谢颂海[1] LU Long-fei;XIE Song-hai(Department of Chemistry,Fudan University,Shanghai 200438,China)

机构地区：[1]复旦大学化学系,上海200438

出　　处：《电子显微学报》2023年第2期216-222,共7页Journal of Chinese Electron Microscopy Society

基　　金：国家自然科学基金资助项目(No.21974027)。

摘　　要：样品漂移是影响透射电子显微镜分辨率的重要因素之一。近年来有一系列关于透射电镜图像漂移校正的算法,如Unblur、MotionCor和MotionCor2等,在冷冻电镜三维重构方面应用效果显著,但无法应用于材料表征电镜通常配备的CCD相机的图像漂移校正。本文提出了一种基于图像分割和盲反卷积的透射电镜图像漂移校正方法,在Goldstein和Fattal提出的利用功率谱的统计不规则性估计卷积核算法的基础上,利用图像分割解决了电镜图像样品部分和背景部分的运动方式不一致问题,能够应用于CCD相机的图像漂移校正。实验结果表明,本文的方法能够有效提高电镜图像的分辨率。The sample drift is one of the important factors that affect the resolution of transmission electron microscopy(TEM).In recent years,there have been a series of algorithms for TEM image drift correction,such as Unblur,MotionCor and MotionCor2.These algorithms have realized remarkable applications in the three-dimensional reconstruction of Cryo-EM.However,they have not been extended to the drift correction of TEM image.In this paper,we proposed a method for drift correction o£TEM image based on image segmentation and blind deconvolution.Based on the estimation method of convolution kernel proposed by Goldstein and Fattal,the image segmentation was used to solve the problem of inconsistencies in the motion mode of the sample part and the background part of TEM image.Our developed method can be applied to the image drift correction of CCD camera.Experimental result show that our proposed method can effectively improve the resolution of TEM images.

关 键 词：透射电子显微镜 图像漂移校正 图像分割 盲反卷积 

分 类 号：TN16[电子电信—物理电子学] TP319[自动化与计算机技术—计算机软件与理论]