Abstract
We present a self-supervised hologram reconstruction neural network trained using a physics-consistency loss, which achieves superior generalization to reconstruct holograms of various samples, without previously having/seeing any experimental data or prior knowledge regarding the samples.
© 2023 The Author(s)
PDF ArticleMore Like This
M. Hossein Eybposh, Aram Moossavi, Jiayi Xu, Nicholas W. Caira, and Nicolas C. Pégard
Th1A.6 Digital Holography and Three-Dimensional Imaging (DH) 2022
Luzhe Huang, Tairan Liu, Xilin Yang, Yi Luo, Yair Rivenson, and Aydogan Ozcan
ATh1D.5 CLEO: Applications and Technology (CLEO:A&T) 2022
Soren Nelson and Rajesh Menon
JTu5A.3 Applied Industrial Spectroscopy (AIS) 2021