Amplitude/Phase Retrieval for Terahertz Holography With Supervised and Unsupervised Physics-Informed Deep Learning
Open Access
- 3 January 2024
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Terahertz Science and Technology
- Vol. 14 (2) , 208-215
Abstract
Most neural networks proposed for computational imaging (CI) in the terahertz (THz) bands require a large amount of experimental data to optimize their weights and biases. However, obtaining a sufficient number of ground-truth images for training is challenging in the THz domain due to the requirements of environmental and system stability, as well as the lengthy data acquisition process. To overcome this limitation, this paper proposes novel supervised and unsupervised physics-informed deep learning (DL) methods for amplitude and phase recovery by incorporating angular spectrum diffraction theory as prior knowledge. Firstly, we demonstrate that our unsupervised dual network can predict both amplitude and phase simultaneously, overcoming the limitation of previous studies that could only predict phase objects. This is demonstrated using synthetic 2D image data as well as measured diffraction images. The advantage of unsupervised DL is its ability to be used directly without labeling by human experts. Additionally, we address supervised DL, which is a concept of general applicability. We introduce training with a database set of 2D images taken in the visible spectra range and numerically modified by us to emulate THz images. This approach allows us to avoid the prohibitively time-consuming collection of a large number of THz-frequency images. Furthermore, we employ a combination method that enhances the sharpness of image edges, improves contrast, and effectively aligns the approach with the ground truth. The results obtained using both approaches represent the initial steps towards fast holographic THz imaging with reference-beam-free, low-cost power detection.Keywords
All Related Versions
Funding Information
- XF-IJRC
- German Research Foundation DFG (RO 770/48-1)
- BMBF
- Samson
This publication has 42 references indexed in Scilit:
- Phase imaging with an untrained neural networkLight: Science & Applications, 2020
- Learning to synthesize: robust phase retrieval at low photon countsLight: Science & Applications, 2020
- Terahertz phase imaging and biomedical applicationsOptics & Laser Technology, 2020
- Terahertz Quality Inspection for Automotive and Aviation IndustriesJournal of Infrared, Millimeter, and Terahertz Waves, 2019
- Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equationsJournal of Computational Physics, 2019
- An equation-of-state-meter of quantum chromodynamics transition from deep learningNature Communications, 2018
- Biomedical Applications of Terahertz Spectroscopy and ImagingTrends in Biotechnology, 2016
- Iterative solution to twin image problem in in-line digital holographyOptics and Lasers in Engineering, 2013
- In Introductory Review to THz Non-Destructive Testing of Composite MaterJournal of Infrared, Millimeter, and Terahertz Waves, 2012
- Diffraction transfer function and its calculation of classic diffraction formulaOptics Communications, 2007