摘要
数字全息术(DH)是监测透明样品定量三维信息的一种重要技术.然而,常规数字全息重建中需要相位畸变补偿和解包裹,严重影响了相位重建速度和重建精度.提出一种融合空洞卷积和注意力机制的改进残差Unet方法,实现了数字全息端到端相位重建,简化成像过程,提高了图像重建质量.此外,该方法还可以通过调整残差块,得到最优的实时重建网络模型.实验结果表明,所提基于深度学习的相位重建方法能够实时获得样品精确的三维形貌信息,有利于对动态样品进行实时监测.
Digital holography(DH)is critical for monitoring quantitative threedimensional information of transparent samples.However,phase aberration compensation and unwrapping are needed in conventional digital holographic reconstruction,which adversely affect its speed and accuracy.We propose an improved residual Unet method that combines dilated convolution and attention mechanism to implement endtoend phase reconstruction of DH,which simplifies the imaging process and improves the quality of image reconstruction.In addition,the proposed method can further optimize the network model for realtime reconstruction by adjusting residual blocks.The experimental results reveal that the proposed phase reconstruction method based on deep learning can obtain accurate threedimensional information of samples in real time,which benefits realtime monitoring for dynamic samples.
作者
李坤格
王华英
柳旭
王杰宇
王文健
杨柳
Li Kunge;Wang Huaying;Liu Xu;Wang Jieyu;Wang Wenjian;Yang Liu(School of Mathematics and Physics Science and Engineering,Hebei University of Engineering,Handan 056038,Hebei,China;Hebei Computational Optical Imaging and Photoelectric Detection Technology Innovation Center,Handan 056038,Hebei,China;Hebei International Joint Research Center for Computational Optical Imaging and Intelligent Sensing,Handan 056038,Hebei,China)
出处
《激光与光电子学进展》
CSCD
北大核心
2023年第6期162-169,共8页
Laser & Optoelectronics Progress
基金
国家自然科学基金面上项目(62175059)
河北省自然科学基金重点项目(2018402285)
河北省创新能力提升计划项目(20540302D)。
关键词
数字全息术
相位重建
深度学习
残差网络
digital holography
phase reconstruction
deep learning
residual network
