On the use of deep learning for computational imaging

By George Barbastathis. NanoScience + Engineering, 2020

574 citations recorded by the source corpus.

Deep learning has emerged as a class of optimization algorithms proven to be effective for a variety of inference and decision tasks. Similar algorithms, with appropriate modifications, have also been widely adopted for computational imaging. Here, we review the basic tenets of deep learning and computational imaging, and overview recent progress in two applications: super resolution and phase retrieval.

Subfields: Atomic and Molecular Physics, and Optics, Geophysics, Radiation

View Paper: https://www.semanticscholar.org/paper/3f08231000088a88abca7bcc7be1075e425b8bc5

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Atomic and Molecular Physics, and Optics

In Geophysics

In Radiation

Citations in the corpus

Learning-based lensless imaging through optically thick scattering media (2019, unknown): “22,23 24 – 32 Horisaki et al. 24 proposed to use the support vector regression architecture to reconstruct the scattering images with a set of human face data for training.”
Inference in artificial intelligence with deep optics and photonics (2020, methodology): “Solving inverse problems for reconstruction and enhancement of microscopy images has been a hot topic in research for decades 78 ; a key component of the previous approaches is the establishment of a forward model of the imaging system 79 .”
Neural holography with camera-in-the-loop training (2020, methodology): “Recent surveys on the use of deep learning in holographic and computational imaging discusses these and other related techniques in detail [Barbastathis et al. 2019; Rivenson et al. 2019].”
Snapshot Compressive Imaging: Theory, Algorithms, and Applications (2021, result): “It has been shown that an encoder–decoder structure with a skip-connection CNN can usually lead to good results for SCI reconstruction, similar to other computational imaging problems [90].”
Deep Learning Techniques for Inverse Problems in Imaging (2020, background): “See [53] for a recent optics-focused review article.”
Far-field super-resolution ghost imaging with a deep neural network constraint (2022, background): “An alternative but increasingly important approach is deep learning that is based on data prior assumptions 25 – 27 .”
Deep learning in optical metrology: a review (2022, methodology): “The reader is referred to the earlier review by Barbastathis et al.(45) for more detailed information on this topic.”