摘要
全息重建方法是全息成像技术和全息术的一个核心部分。本文提出了一种基于U_net网络的单幅全息图重建方法。首先使用基于马赫-增德尔干涉仪的光路结构全息实验成像系统记录全息图,将得到的全息图作为训练集,然后采用训练集对U_net网络进行训练,在经过训练后得到全息图的数学模型,用该数学模型便可以对全息图进行重建。仿真和实验结果表明,基于U_net网络的单幅全息图重建方法计算速度更快还只需要更少的测量数据,并且仅使用单幅全息图就可以重建图像,且重建图像没有直流项和孪生项的干扰。
The holographic reconstruction method is a core part of holographic imaging technology and holography.Firstly,the hologram is recorded by the optical path structure holographic experimental imaging system based on Mach-Zehnder interferometer. The obtained hologram is used as the training set,then the training set is trained by U_net network,and the mathematical model of hologram is obtained after training. The hologram can be reconstructed with this mathematical model.Simulation and experimental results show that the single hologram reconstruction method based on U_net network can calculate faster and only need less measurement data,and only use a single hologram to reconstruct the image,and the reconstructed image has no DC and twin items.
作者
李菊
李军
LI Ju;LI Jun(School of Physics and Telecommunications Engineering,South China Normal University,Guangzhou 51006,China)
出处
《激光杂志》
北大核心
2020年第1期96-99,共4页
Laser Journal
基金
广东省自然科学基金(No.2016A010101021)
广州市科技计划(No.201607010275)
关键词
全息重建
U_net网络
单幅全息
数学模型
holographic reconstruction
U_net Network
single hologram
mathematical model
