Improved Machine Learning Approach for Wavefront Sensing

Guo, Hongyang; Xu, Yangsheng; Li, Qing; Du, Shengping; He, Dong; Wang, Qiang; Huang, Yongmei

doi:10.3390/s19163533

Cited by 39 publications

(21 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Artificial neural networks [ 28 , 29 , 30 ], which belong to machine learning, are input–output information processors composed of parallel layers of elements or neurons, loosely modeled on biological neurons, which possess local memory and are capable of elementary arithmetic. They can be used to learn and store a great deal of nonlinear mapping relations from the input–output model.…”

Section: Piston Error Detection Methods Using Bp Artificial Neural Networkmentioning

confidence: 99%

Piston Error Measurement for Segmented Telescopes with an Artificial Neural Network

Yue

2021

Sensors

View full text Add to dashboard Cite

A piston error detection method is proposed based on the broadband intensity distribution on the image plane using a back-propagation (BP) artificial neural network. By setting a mask with a sparse circular clear multi-subaperture configuration in the exit pupil plane of a segmented telescope to fragment the pupil, the relation between the piston error of segments and amplitude of the modulation transfer function (MTF) sidelobes is strictly derived according to the Fourier optics principle. Then the BP artificial neural network is utilized to establish the mapping relation between them, where the amplitudes of the MTF sidelobes directly calculated from theoretical relationship and the introduced piston errors are used as inputs and outputs respectively to train the network. With the well trained-network, the piston errors are measured to a good precision using one in-focused broadband image without defocus division as input, and the capture range achieving the coherence length of the broadband light is available. Adequate simulations demonstrate the effectiveness and accuracy of the proposed method; the results show that the trained network has high measurement accuracy, wide detection range, quite good noise immunity and generalization ability. This method provides a feasible and easily implemented way to measure piston error and can simultaneously detect the multiple piston errors of the entire aperture of the segmented telescope.

show abstract

Section: Piston Error Detection Methods Using Bp Artificial Neural Networkmentioning

confidence: 99%

Piston Error Measurement for Segmented Telescopes with an Artificial Neural Network

Yue

2021

Sensors

View full text Add to dashboard Cite

show abstract

“…1C. Dual-view data were deconvolved after export from MicroManager using custom fusion and deconvolution software (Guo et al, 2019). All datasets were then screened and subsequently analyzed.…”

Section: Image Analysismentioning

confidence: 99%

Multi-tissue patterning drives anterior morphogenesis of theC. elegansembryo

Grimbert

Mastronardi

Christensen

et al. 2020

Preprint

View full text Add to dashboard Cite

Complex structures derived from multiple tissue types are challenging to study in vivo, and our knowledge of how cells from different tissues are coordinated is limited. Model organisms have proven invaluable for improving our understanding of how chemical and mechanical cues between cells from two different tissues can govern specific morphogenetic events. Here we usedCaenorhabditis elegans as a model system to show how cells from three different tissues are coordinated to give rise to the anterior lumen. This poorly understood process has remained a black box for embryonic morphogenesis. Using various microscopy and software approaches, we describe the movements and patterns of epidermal cells, neuroblasts and pharyngeal cells that contribute to lumen formation. The anterior-most pharyngeal cells (arcade cells) may provide the first marker for the location of the future lumen and facilitate the patterning of the surrounding neuroblasts. These neuroblast patterns control the rate of migration of the anterior epidermal cells, whereas the epidermal cells ultimately reinforce and control the position of the future lumen, as they must join with the pharyngeal cells for their epithelialization. Our studies are the first to characterize anterior morphogenesis in C. elegans in detail and should lay the framework for identifying how these different patterns are controlled at the molecular level.

show abstract

“…8 There has been some limited work related to image-based wavefront sensing. [9][10][11][12] A recent paper 13 deals with some of the same topics as the present publication but focuses on laser communication with monochromatic light and does not touch upon the issues of phase screen generation, noise, blurring by wide-band imaging, closed-loop operation, and bit depth, which all are important for our application. Another recent article 14 briefly describes an interesting image sharpening approach using a neural network based on unsupervised learning.…”

Section: Introductionmentioning

confidence: 99%

Image-based wavefront sensing for astronomy using neural networks

Andersen

Owner-Petersen

Enmark

2020

J. Astron. Telesc. Instrum. Syst.

View full text Add to dashboard Cite

Motivated by the potential of nondiffraction limited, real-time computational image sharpening with neural networks in astronomical telescopes, we studied wavefront sensing with convolutional neural networks based on a pair of in-focus and out-of-focus point spread functions. By simulation, we generated a large dataset for training and validation of neural networks and trained several networks to estimate Zernike polynomial approximations for the incoming wavefront. We included the effect of noise, guide star magnitude, blurring by wide-band imaging, and bit depth. We conclude that the "ResNet" works well for our purpose, with a wavefront RMS error of 130 nm for r 0 ¼ 0.3 m, guide star magnitudes 4 to 8, and inference time of 8 ms. It can also be applied for closed-loop operation in an adaptive optics system. We also studied the possible use of a Kalman filter or a recurrent neural network and found that they were not beneficial to the performance of our wavefront sensor. © The Authors. Published by SPIE under a Creative Commons Attribution 4.0 Unported License. Distribution or reproduction of this work in whole or in part requires full attribution of the original publication, including its DOI.

show abstract

Improved Machine Learning Approach for Wavefront Sensing

Cited by 39 publications

References 24 publications

Piston Error Measurement for Segmented Telescopes with an Artificial Neural Network

Piston Error Measurement for Segmented Telescopes with an Artificial Neural Network

Multi-tissue patterning drives anterior morphogenesis of theC. elegansembryo

Image-based wavefront sensing for astronomy using neural networks

Contact Info

Product

Resources

About