Abstract Deep learning neural networks are used for wavefront sensing and aberration correction in atmospheric turbulence without any sensor (i.e. reconstruction of the phase from distorted image object). We compared found characteristics direct indirect ways: (i) directly reconstructing phase; (ii) Zernike coefficients then calculating phase. verified generalization ability performance network a single object multiple objects. What’s more, we effect pool feasibility real environment.

The time-delay problem, which is introduced by the response time of hardware for correction, a critical and non-ignorable problem adaptive optics (AO) systems. It will result in significant wavefront correction errors while turbulence changes severely or system responses slowly. Predictive AO proposed to alleviate more accurate stable corrections real time-varying atmosphere. However, existing prediction approaches either lack ability extract non-linear temporal features, overlook...

Digital holographic microscopy enables the measurement of quantitative light field information and visualization transparent specimens. It can be implemented for complex amplitude imaging thus investigation biological samples including tissues, dry mass, membrane fluctuation, etc. Currently, deep learning technologies are developing rapidly have already been applied to various important tasks in coherent imaging. In this paper, an optimized structural convolution neural network PhaseNet is...

This paper presents a new, yet simple and effective approach to modeling industrial Radio Frequency heating systems, using the wave equation applied in three dimensions instead of conventional electrostatics method. The central idea is that tank oscillatory circuit excited an external source. then excites applicator which used heat or dry processed load. Good agreement was obtained between experimental numerical data, namely S11-parameter, phase, patterns for different sized loads positions.

Adaptive optics (AO) has great applications in many fields and attracted wide attention from researchers. However, both traditional deep learning-based AO methods have inherent time delay caused by wavefront sensors controllers, leading to the inability truly achieve real-time atmospheric turbulence correction. Hence, future turbulent prediction plays a particularly important role AO. Facing challenge of accurately predicting stochastic turbulence, we combine convolutional neural network...

Microwave heating plays an important role in the processing & of foods food industry and at home, but applicator design is frequently arduous task. Thefinite-difference time-domain (FDTD) method can be used to model complex geometries guide different types applicators. In reported research, FDTD approach has been employed characterize electric field distribution a rectangular waveguide applicator, carrying TE10-mode aperture distribution, terminated oversize waveguide/cavity operating 915...

Abstract Online partial discharge (PD) detection still remains a very challenging task because of the strong electromagnetic interferences. In this paper, new method de‐noising, using complex Daubechies wavelet (CDW) transform, has been proposed. It is relatively recent enhancement to real‐valued transform two important properties, which are nearly shift invariant and availability phase information. Those properties give CDW superiority over other transform. On basis threshold algorithm...

Surface plasmon resonance microscopy (SPRM) has been massively applied for near-field optical measurement, sensing, and imaging because of its high detection sensitivity, nondestructive, noninvasive, wide-field, label-free capabilities. However, the transverse propagation characteristic surface wave generated during (SPR) leads to notable “tail” patterns in SPR image, which severely deteriorates image quality. Here, we propose an incidence angle scanning method SPRM obtain a with exceptional...

The optical sparse aperture technique can improve the imaging resolution significantly under ideal co-phase condition. However, position deviation between different sub-apertures leads to notorious errors, seriously impacting image quality. While arises in practical applications, it is difficult detect errors real-time for traditional iterative algorithms because of their narrow detection range and long-time iteration process. deep neural network has shown its potential information process,...

Deep learning technology has shown excellent performances and successful applications in optical information processing. However, the long-time training, large amount of manually labeled data generalization capability hinder application deep neural network (DNN) under supervised learning. The image prior (DIP) opinion promotes development untrained network, which can learn from one image. Here we propose a DIP-based strategy to nest DNN into physical model for finding optimal solution...

Deep learning has recently shown great potential in computational imaging. Here, we propose a deep-learning-based reconstruction method to realize the sparse-view imaging of fiber internal structure holographic diffraction tomography. By taking sinogram as input and cross-section image obtained by dense-view ground truth, neural network can reconstruct from sinogram. It performs better than corresponding filtered back-projection algorithm with sinogram, both case simulated data real...

Deep learning techniques can be introduced into the digital holography to suppress coherent noise. It is often necessary first make a dataset of noisy and noise-free phase images train network. However, are difficult obtain in practical holographic applications. Here we propose label-free training algorithms based on self-supervised learning. A dilated blind spot network built learn from real noise level function estimate function. Then they trained together via maximizing constrained...

The working temperature of transformer oil are different in regions and seasons. In this paper, the partial discharge (PD) charged metal particles flowing under were studied. φ-U-N diagram PD characteristic constructed, amplitude number gotten. Metal particle trajectories simulated using finite element method temperature. Results show that when is below 343.15 K, reduce gradually with increasing. However, to begins increase. phase mainly concentrated vicinity 90° 270 When from 333.15 K...

We propose a model-enhanced network with unpaired single-shot data for solving the imaging blur problem of an optical sparse aperture (OSA) system. With only one degraded image captured from system and "arbitrarily" selected clear image, cascaded neural is iteratively trained denoising restoration. computational degradation model enhancement, our method able to improve contrast, restore blur, suppress noise images in simulation experiment. It can achieve better restoration performance fewer...