Phase Retrieval Using Conditional Generative Adversarial Networks

Uelwer, Tobias; Oberstrab, Alexander; Harmeling, Stefan

doi:10.1109/icpr48806.2021.9412523

Cited by 17 publications

(19 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In [16], similar concepts were used, namely, a generator was trained to take as input the Fourier magnitude and output a predicted image. The generator was trained with a linear combination of conditional and advesarial losses.…”

Section: Overview Of Recent Deep Learning Based Methodsmentioning

confidence: 99%

PR-DAD: Phase Retrieval Using Deep Auto-Decoders

Gugel¹,

Dekel²

2022

Preprint

View full text Add to dashboard Cite

Phase retrieval is a well known ill-posed inverse problem where one tries to recover images given only the magnitude values of their Fourier transform as input. In recent years, new algorithms based on deep learning have been proposed, providing breakthrough results that surpass the results of the classical methods. In this work we provide a novel deep learning architecture PR-DAD (Phase Retrieval Using Deep Auto-Decoders), whose components are carefully designed based on mathematical modeling of the phase retrieval problem. The architecture provides experimental results that surpass all current results.

show abstract

Section: Overview Of Recent Deep Learning Based Methodsmentioning

confidence: 99%

PR-DAD: Phase Retrieval Using Deep Auto-Decoders

Gugel¹,

Dekel²

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…The L MAE is used for the Fashion-MNIST dataset. Furthermore, we report the results of an MLP trained with an adversarial loss in combination with L MAE (PRCGAN) as proposed in [21]. For our proposed CPR network we consider a cascade of five MLPs with three hidden layers where we increased the scales of the (intermediate) reconstructions according to Tab.…”

Section: Methodsmentioning

confidence: 99%

“…Another class of learned methods rely on the optimization of a latent variable of a learned generative model [7,21] and produce high quality results. However, these methods require a training phase and an optimization phase during application and are therefore very costly.…”

Section: Related Workmentioning

confidence: 99%

“…In this section, we empirically evaluate the performance of our model. In order to do this, we report the results of the fully-convolutional residual network (ResNet) employed by Nishizaki et al [16], the multi-layer-perceptron (MLP) used in [21] and the PRCGAN [21]. In addition to these learned networks we include the results of the well-established HIO algorithm [4] and the RAAR algorithm [12] as a baseline.…”

Section: Experimental Evaluationmentioning

confidence: 99%

See 1 more Smart Citation

Non-Iterative Phase Retrieval With Cascaded Neural Networks

Uelwer¹,

Hoffmann²,

Harmeling³

2021

Preprint

Self Cite

View full text Add to dashboard Cite

Fourier phase retrieval is the problem of reconstructing a signal given only the magnitude of its Fourier transformation. Optimizationbased approaches, like the well-established Gerchberg-Saxton or the hybrid input output algorithm, struggle at reconstructing images from magnitudes that are not oversampled. This motivates the application of learned methods, which allow reconstruction from non-oversampled magnitude measurements after a learning phase. In this paper, we want to push the limits of these learned methods by means of a deep neural network cascade that reconstructs the image successively on different resolutions from its non-oversampled Fourier magnitude. We evaluate our method on four different datasets (MNIST, EMNIST, Fashion-MNIST, and KMNIST) and demonstrate that it yields improved performance over other non-iterative methods and optimization-based methods.

show abstract

“…Using pre-trained Gaussian denoisers and iterative algorithms, Deep-prior-based sparse representation (Shi et al, 2020), prDeep (Metzler et al, 2018) and Deep-ITA (Wang et al, 2020b), are solutions robust to noise in the case where the corruption is small enough to be approximately Gaussian. The end-to-end solution investigated by Uelwer et al (2019) features some robustness to Poisson shot noise, but struggles to generalize to complicated datasets. Meanwhile, the "physics-informed" (2020) tested the deep decoder, both untrained neural networks, for Fourier phase retrieval, but did not consider the holographic setting.…”

Section: Related Workmentioning

confidence: 99%

Phase Retrieval with Holography and Untrained Priors: Tackling the Challenges of Low-Photon Nanoscale Imaging

Lawrence,

Barmherzig,

et al. 2020

Preprint

View full text Add to dashboard Cite

Phase retrieval is the inverse problem of recovering a signal from magnitude-only Fourier measurements, and underlies numerous imaging modalities, such as Coherent Diffraction Imaging (CDI). A variant of this setup, known as holography, includes a reference object that is placed adjacent to the specimen of interest before measurements are collected. The resulting inverse problem, known as holographic phase retrieval, is well-known to have improved problem conditioning relative to the original. This innovation, i.e. Holographic CDI, becomes crucial at the nanoscale, where imaging specimens such as viruses, proteins, and crystals require low-photon measurements. This data is highly corrupted by Poisson shot noise, and often lacks low-frequency content as well. In this work, we introduce a dataset-free deep learning framework for holographic phase retrieval adapted to these challenges. The key ingredients of our approach are the explicit and flexible incorporation of the physical forward model into an automatic differentiation procedure, the Poisson log-likelihood objective function, and an optional untrained deep image prior. We perform extensive evaluation under realistic conditions. Compared to competing classical methods, our method recovers signal from higher noise levels and is more resilient to suboptimal reference design, as well as to large missing regions of low frequencies in the observations. To the best of our knowledge, this is the first work to consider a dataset-free machine learning approach for holographic phase retrieval.

show abstract

Phase Retrieval Using Conditional Generative Adversarial Networks

Cited by 17 publications

References 21 publications

PR-DAD: Phase Retrieval Using Deep Auto-Decoders

PR-DAD: Phase Retrieval Using Deep Auto-Decoders

Non-Iterative Phase Retrieval With Cascaded Neural Networks

Phase Retrieval with Holography and Untrained Priors: Tackling the Challenges of Low-Photon Nanoscale Imaging

Contact Info

Product

Resources

About