Microscopic imaging providing the real-space information of matter, plays an important role for understanding the correlations between structure and properties in the field of materials science. For the microscopic images of different kinds of objects at different scales, it is a time-consuming task to retrieve useful information on morphology, size, distribution, intensity etc. Alternatively, deep learning has shown great potential in the applications on complicated systems for its ability of extracting useful information automatically. Recently, researchers have utilized deep learning methods on imaging analysis to identify structures and retrieve the linkage between microstructure and performance. In this review, we summarize the recent progresses of the applications of deep learning analysis on microscopic imaging, including scanning electron microscopy (SEM), transmission electron microscopy (TEM), and scanning probe microscopy (SPM). We present sequentially the basic concepts of deep learning methods, the review of the applications on imaging analysis, and our perspective on the future development. Based on the published results, a general workflow of deep learning analysis is put forward.

显微成像提供了物质的真实空间信息，对于理解材料科学领域中结构与特性之间的相关性起着重要作用。对于不同比例的不同种类的物体的显微图像，检索形态，大小，分布，强度等有用信息是一项耗时的任务。或者，深度学习在复杂系统中的应用显示出巨大的潜力。自动提取有用信息的能力。最近，研究人员已将深度学习方法用于成像分析，以识别结构并检索微观结构与性能之间的联系。在这篇综述中，我们总结了深度学习分析在显微成像中的应用的最新进展，包括扫描电子显微镜（SEM），透射电子显微镜（TEM）和扫描探针显微镜（SPM）。我们依次介绍了深度学习方法的基本概念，图像分析应用程序的回顾以及对未来发展的看法。基于已发布的结果，提出了深度学习分析的一般工作流程。