基于自监督学习实现电子显微图像降噪

姚家豪,丁洋,国洪轩*,孙立涛*

基于自监督学习实现电子显微图像降噪

姚家豪,丁洋,国洪轩*,孙立涛*

(MEMS教育部重点实验室,东南大学电子科学与工程学院,江苏 南京 210096)

摘 要  在电子显微镜的表征工作中,噪声可以掩盖或者干扰有用的信号并对后续研究造成不可忽视的影响。本文发展了一种使用自监督深度学习技术对电子显微图像进行降噪的新方法,该方法搭建了基于U-Net的新型降噪神经网络模型,利用最大模糊池化以及注意力机制提高降噪能力。最后,本研究通过多种电子显微实验数据验证了所提出方法的有效性。相比有监督学习,本方法更适合难以获得干净数据的电子显微图像场景,此外本方法比传统机器学习拥有更好的降噪效果和效率。

关键词  电子显微图像;自监督学习;神经网络;图像降噪

中图分类号:TP391.41;TP18;TN16   文献标识码A   doi:10.3969/j.issn.1000-6281.2024.01.010

 

Design and deformation mechanism of CoCrNi medium entropy alloy with gradient heterostructure

LUO Xian-min1,SU Hong-hong1,MAO Sheng-cheng1*,ZHANG Ze2,HAN Xiao-dong1

(1.Faculty of Materials and Manufacturing, Beijing University of Technology, Beijing 100124;2. School of Materials Science and Engineering, Zhejiang University, Hangzhou Zhejinag, 310027, China)

Abstract   CoCrNi is a typical medium entropy alloy with high plasticity but low tensile strength. The trade-off problem limits its engineering applications. In this work, the heterogeneous structure of the gradient grain and twin was introduced into a CrCoNi medium entropy alloy by rolling annealing and rotating accelerated shot peening. The results show that this design achieves an excellent strong plasticity with the yield strength, tensile strength and fracture plasticity of 665 MPa, 950 MPa and 40.6%, respectively. During tensile deformation, the gradient heterostructure can produce the hetero deformation-induced hardening, improve the yield strength of the alloy, and maintain a high work hardening rate and an excellent ductility. Therefore, the introduction of the gradient heterostructure based on rolling, annealing, and rotating shot peening is an effective way to overcome the trade-off between strength and ductility.

Keywords   CoCrNi medium entropy alloy;gradient heterostructure;hetero-deformation induced hardenging

 

“全文下载请到同方知网,万方数据库或重庆维普等数据库中下载!”