Deep learning for digital holography: a review

Zeng, Tianjiao; Zhu, Youhua; Lam, Edmund Y.

doi:10.1364/oe.443367

Cited by 114 publications

(38 citation statements)

References 98 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Additionally, the use of deep learning technologies allows the elimination of the restored image of the 0th order; it compensates for phase aberrations and suppresses noise (for example, with a Kalman filter) [ 50 ].…”

Section: Methods and Algorithms For Processing And Restoring Of Holog...mentioning

confidence: 99%

Advances in Digital Holographic Interferometry

et al. 2022

View full text Add to dashboard Cite

Holographic interferometry is a well-established field of science and optical engineering. It has a half-century history of successful implementation as the solution to numerous technical tasks and problems. However, fast progress in digital and computer holography has promoted it to a new level of possibilities and has opened brand new fields of its application. In this review paper, we consider some such new techniques and applications.

show abstract

Section: Methods and Algorithms For Processing And Restoring Of Holog...mentioning

confidence: 99%

Advances in Digital Holographic Interferometry

et al. 2022

View full text Add to dashboard Cite

show abstract

“…The window size has been fixed to 4 with an initial patch size of 4x4. TViT and TSwinT contrasts with canonical ViT architectures as these models are usually able to learn high-quality intermediate representations with large amounts of data as described in [41] and [22]. TVGG is introduced to reduce the number of parameters of the original VGG16 [37] architecture for comparison purposes.…”

Section: Tiny Network: Tvit Tswint and Tvggmentioning

confidence: 99%

“…Contactless sensors are thus desired to control 3D motions with a high In this paper, we aim to illustrate a new high-profile application of machine learning by elevating DHM and autofocusing to a new level. Whereas many studies focus on life science microscopy [18,32,22], this work explores extended visual capabilities offered by combining DH and last generation of DL algorithms such as Vision Transformer (ViT) [33] and Swin-Transformer (SwinT) [34] networks for applications in micro-robotics [12,13] or in real-time 3D microscopy [32]. We introduce for the first time the neural network Transformer architectures applied in optics and advanced coherent imaging field, such as digital holography.…”

Section: Introductionmentioning

confidence: 99%

Fast Autofocusing using Tiny Transformer Networks for Digital Holographic Microscopy

Cuenat,

Andréoli,

André

et al. 2022

Preprint

View full text Add to dashboard Cite

The numerical wavefront backpropagation principle of digital holography confers unique extended focus capabilities, without mechanical displacements along z-axis. However, the determination of the correct focusing distance is a non-trivial and time consuming issue. A deep learning (DL) solution is proposed to cast the autofocusing as a regression problem and tested over both experimental and simulated holograms. Single wavelength digital holograms were recorded by a Digital Holographic Microscope (DHM) with a 10x microscope objective from a patterned target moving in 3D over an axial range of 92 µm. Tiny DL models are proposed and compared such as a tiny Vision Transformer (TViT), tiny VGG16 (TVGG) and a tiny Swin-Transfomer (TSwinT). The experiments show that the predicted focusing distance Z Pred R is accurately inferred with an accuracy of 1.2 µm in average in comparison with the DHM depth of field of 15 µm. Numerical simulations show that all tiny models give the Z Pred R with an error below 0.3 µm. Such a prospect would significantly improve the current capabilities of computer vision position sensing in applications such as 3D microscopy for life sciences or micro-robotics. Moreover, all models reach state of the art inference time on CPU, less than 25 ms per inference.

show abstract

“…In contrast to the conventional phase recovery algorithms that mainly rely on theoretical knowledge and phase propagation models, supervised DL methods often use large-scale datasets for training a black-box model to solve the inverse problem numerically. Therefore, prior knowledge about the propagation model and the system parameters is not necessary to construct DL networks [8].…”

Section: Related Work On Deep Learning-based Dhmentioning

confidence: 99%

“…Recovering the object wave from the captured hologram H(x, y) facilitates 3D-shape reconstruction, due to the linear relationship between the object thickness and incurred phase shift [8]. Therefore, we need to eliminate the zero-order terms |O(x, y)| 2 and |R(x, y)| 2 and the noise term using filtering methods before the phase recovery.…”

Section: Introductionmentioning

confidence: 99%

DH-GAN: A Physics-driven Untrained Generative Adversarial Network for 3D Microscopic Imaging using Digital Holography

Chen¹,

Wang²,

Razi³

et al. 2022

Preprint

View full text Add to dashboard Cite

Digital holography is a 3D imaging technique by emitting a laser beam with a plane wavefront to an object and measuring the intensity of the diffracted waveform, called holograms. The object's 3D shape can be obtained by numerical analysis of the captured holograms and recovering the incurred phase. Recently, deep learning (DL) methods have been used for more accurate holographic processing. However, most supervised methods require large datasets to train the model, which is rarely available in most DH applications due to the scarcity of samples or privacy concerns. A few one-shot DL-based recovery methods exist with no reliance on large datasets of paired images. Still, most of these methods often neglect the underlying physics law that governs wave propagation. These methods offer a black-box operation, which is not explainable, generalizable, and transferrable to other samples and applications. In this work, we propose a new DL architecture based on generative adversarial networks that uses a discriminative network for realizing a semantic measure for reconstruction quality while using a generative network as a function approximator to model the inverse of hologram formation. We impose smoothness on the background part of the recovered image using a progressive masking module powered by simulated annealing to enhance the reconstruction quality. The proposed method is one of its kind that exhibits high transferability to similar samples, which facilitates its fast deployment in time-sensitive applications without the need for retraining the network. The results show a considerable improvement to competitor methods in reconstruction quality (about 5 dB PSNR gain) and robustness to noise (about 50% reduction in PSNR vs noise increase rate).

show abstract

Deep learning for digital holography: a review

Cited by 114 publications

References 98 publications

Advances in Digital Holographic Interferometry

Advances in Digital Holographic Interferometry

Fast Autofocusing using Tiny Transformer Networks for Digital Holographic Microscopy

DH-GAN: A Physics-driven Untrained Generative Adversarial Network for 3D Microscopic Imaging using Digital Holography

Contact Info

Product

Resources

About