Toward simple, generalizable neural networks with universal training for low-SWaP hybrid vision

Author(s):

Muminov, Baurzhan; Perry, Altai; Hyder, Rakib; Asif, M. Salman & Vuong, Luat T.

Abstract:

“Speed, generalizability, and robustness are fundamental issues for building lightweight computational cameras. Here we demonstrate generalizable image reconstruction with the simplest of hybrid machine vision systems: linear optical preprocessors combined with no-hidden-layer, “small-brain” neural networks. Surprisingly, such simple neural networks are capable of learning the image reconstruction from a range of coded diffraction patterns using two masks. We investigate the possibility of generalized or “universal training” with these small brains. Neural networks trained with sinusoidal or random patterns uniformly distribute errors around a reconstructed image, whereas models trained with a combination of sharp and curved shapes (the phase pattern of optical vortices) reconstruct edges more boldly. We illustrate variable convergence of these simple neural networks and relate learnability of an image to its singular value decomposition entropy of the image. We also provide heuristic experimental results. With thresholding, we achieve robust reconstruction of various disjoint datasets. Our work is favorable for future real-time low size, weight, and power hybrid vision: we reconstruct images on a 15 W laptop CPU with 15,000 frames per second: faster by a factor of 3 than previously reported results and 3 orders of magnitude faster than convolutional neural networks.”

Link to Publications Page

Publication: Photonics Research
Issue/Year: Photonics Research, Volume 9; Number 7; Pages B253; 2021
DOI: 10.1364/prj.416614

Super-resolution imaging by optical incoherent synthetic aperture with one channel at a time

Author(s):

Bulbul, Angika & Rosen, Joseph

Abstract:

“Imaging with an optical incoherent synthetic aperture (SA) means that the incoherent light from observed objects is processed over time from various points of view to obtain a resolution equivalent to single-shot imaging by the SA larger than the actual physical aperture. The operation of such systems has always been based on two-wave interference where the beams propagate through two separate channels. This limitation of two channels at a time is removed in the present study with the proposed SA where the two beams pass through the same single channel at any given time. The system is based on a newly developed self-interference technique named coded aperture correlation holography. At any given time, the recorded intensity is obtained from interference between two waves co-propagating through the same physical channel. One wave oriented in a particular polarization is modulated by a pseudorandom coded phase mask and the other one oriented orthogonally passes through an open subaperture. Both subapertures are multiplexed at the same physical window. The system is calibrated by a point spread hologram synthesized from the responses of a guide star. All the measurements are digitally processed to achieve a final image with a resolution higher than that obtained by the limited physical aperture. This unique configuration can offer alternatives for the current cumbersome systems composed of far apart optical channels in the large optical astronomical interferometers. Furthermore, the proposed concept paves the way to an SA system with a single less-expensive compact light collector in an incoherent optical regime that may be utilized for future ground-based or space telescopes.”

Link to Publications Page

Publication: Photonics Research
Issue/Year: Photonics Research, Volume 9; Number 7; Pages 1172; 2021
DOI: 10.1364/prj.422381

Single-shot digital multiplexed holography for the measurement of deep shapes

Author(s):

Kozacki, Tomasz; Mikuła-Zdańkowska, Marta; Martinez-Carranza, Juan & Idicula, Moncy Sajeev

Abstract:

“This work develops a single-shot holographic profilometer that enables shape characterization of discontinuous deep surfaces. This is achieved by combining hologram frequency multiplexing and an illumination technique of complex amplitude in multi-incidence angle profilometer. Object illumination is carried out from seven directions simultaneously, where the radial angular coordinates of illumination plane waves obey the geometric series. It is shown that: (i) the illumination pattern provides the required frequency separation of all object wavefronts in transverse frequency space, which is necessary for hologram demultiplexing, and (ii) numerical generation of longitudinal scanning function (LSF) is possible, which has large measurement range, high axial resolution, and small side lobes. Low side lobes of LSF and the developed multiplexed field dependent aberration compensation method are essential to minimize the negative influence of speckle noise of single-shot capture on the measurement result. The utility of the proposed method is demonstrated with experimental measurement of heights of two step-like objects.”

Link to Publications Page

Publication: Optics Express
Issue/Year: Optics Express, Volume 29; Number 14; Pages 21965; 2021
DOI: 10.1364/oe.428419

768-ary Laguerre-Gaussian-mode shift keying free-space optical communication based on convolutional neural networks

Author(s):

Luan, Haitao; Lin, Dajun; Li, Keyao; Meng, Weijia; Gu, Min & Fang, Xinyuan

Abstract:

“Beyond orbital angular momentum of Laguerre-Gaussian (LG) modes, the radial index can also be exploited as information channel in free-space optical (FSO) communication to extend the communication capacity, resulting in the LG- shift keying (LG-SK) FSO communications. However, the recognition of radial index is critical and tough when the superposed high-order LG modes are disturbed by the atmospheric turbulences (ATs). In this paper, the convolutional neural network (CNN) is utilized to recognize both the azimuthal and radial index of superposed LG modes. We experimentally demonstrate the application of CNN model in a 10-meter 768-ary LG-SK FSO communication system at the AT \(C^2_n= 10^{-14}m^{-\frac{2}{3}}\). Based on the high recognition accuracy of the CNN model (>95%) in the scheme, a colorful image can be transmitted and the peak signal-to-noise ratio of the received image can exceed 35 dB. We anticipate that our results can stimulate further researches on the utilization of the potential applications of LG modes with non-zero radial index based on the artificial-intelligence-enhanced optoelectronic systems.”

Link to Publications Page

Publication: Optics Express
Issue/Year: Optics Express, Volume 29; Number 13; Pages 19807; 2021
DOI: 10.1364/oe.420176

DL-SI-DHM: a deep network generating the high-resolution phase and amplitude images from wide-field images

Author(s):

Meng, Zhang; Pedrini, Giancarlo; Lv, Xiaoxu; Ma, Jun; Nie, Shouping & Yuan, Caojin

Abstract:

“Structured illumination digital holographic microscopy (SI-DHM) is a high-resolution, label-free technique enabling us to image unstained biological samples. SI-DHM has high re- quirements on the stability of the experimental setup and needs long exposure time. Furthermore, image synthesizing and phase correcting in the reconstruction process are both challenging tasks. We propose a deep-learning-based method called DL-SI-DHM to improve the recording, the reconstruction efficiency and the accuracy of SI-DHM and to provide high-resolution phase imaging. In the training process, high-resolution amplitude and phase images obtained by phase-shifting SI-DHM together with wide-field amplitudes are used as inputs of DL-SI-DHM. The well-trained network can reconstruct both the high-resolution amplitude and phase images from a single wide-field amplitude image. Compared with the traditional SI-DHM, this method significantly shortens the recording time and simplifies the reconstruction process and complex phase correction, and frequency synthesizing are not required anymore. By comparsion, with other learning-based reconstruction schemes, the proposed network has better response to high frequencies. The possibility of using the proposed method for the investigation of different biological samples has been experimentally verified, and the low-noise characteristics were also proved.”

Link to Publications Page

Publication: Optics Express
Issue/Year: Optics Express, Volume 29; Number 13; Pages 19247; 2021
DOI: 10.1364/oe.424718

Caustic Interpretation of the Abruptly Autofocusing Vortex beams

Author(s):

Xiao, Na; Xie, Chen; Jia, Erse; Li, Jiaying; Giust, Remo; Courvoisier, François & Hu, Minglie

Abstract:

“We propose an effective scheme to interpret the abruptly autofocusing vortex beam. In our scheme, a set of analytical formulae are deduced to well predict not only the global caustic, before and after the focal plane, but also the focusing properties of the abruptly autofocusing vortex beam, including the axial position as well as the diameter of focal ring. Our analytical results are in excellent agreement with both numerical simulation and experimental results. Besides, we apply our analytical technique to the fine manipulation of the focusing properties with a scaling factor. This set of methods would be beneficial to a broad range of applications such as particle trapping and micromachinings.”

Link to Publications Page

Publication: Optics Express
Issue/Year: Optics Express, Volume 29; Number 13; Pages 19975; 2021
DOI: 10.1364/oe.430497

Programmable 147 Tb/s (92 Gb/s xhspace0.167em16) optical wireless broadcasting system empowered by a single spatial light modulator and a modified RSS algorithm

Author(s):

You, Quan; Li, Chao; Xiao, Xi & Yu, Shaohua

Abstract:

“We have proposed and experimentally demonstrated a programmable multi-access-point optical wireless broadcasting system with ±15° field-of-view by employing a single spatial light modulator (SLM) and a modified rotated-splitting-SLM algorithm. The 16 access points are generated and arbitrarily distributed by the proposed continuous tunable broadcasting algorithm. The optical beams for each point carry 92-Gb/s PAM-4 optical signal and transmitted over 1 km standard single mode fiber and 1.2 m indoor free space distance, offering a total wireless capacity beyond 1.47 Tb/s. The measured results show that the proposed multi-access-points transmission system with ultra-high transmission capacity and reconfigurability can be used for future indoor wireless mobile networks.”

Link to Publications Page

Publication: Optics Express
Issue/Year: Optics Express, Volume 29; Number 13; Pages 19373; 2021
DOI: 10.1364/oe.428752

Optical vortex lattice: an exploitation of orbital angular momentum

Author(s):

Zhu, Liuhao; Tang, Miaomiao; Li, Hehe; Tai, Yuping & Li, Xinzhong

Abstract:

“Generally, an optical vortex lattice (OVL) is generated via the superposition of two specific vortex beams. Thus far, OVL has been successfully employed to trap atoms via the dark cores. The topological charge (TC) on each optical vortex (OV) in the lattice is only ±1. Consequently, the orbital angular momentum (OAM) on the lattice is ignored. To expand the potential applications, it is necessary to rediscover and exploit OAM. Here we propose a novel high-order OVL (HO-OVL) that combines the phase multiplication and the arbitrary mode-controllable techniques. TC on each OV in the lattice is up to 51, which generates sufficient OAM to manipulate microparticles. Thereafter, the entire lattice can be modulated to desirable arbitrary modes. Finally, yeast cells are trapped and rotated by the proposed HO-OVL. To the best of our knowledge, this is the first realization of the complex motion of microparticles via OVL. Thus, this work successfully exploits OAM on OVL, thereby revealing potential applications in particle manipulation and optical tweezers.”

Link to Publications Page

Publication: Nanophotonics
Issue/Year: Nanophotonics, Volume 10; Number 9; Pages 2487–2496; 2021
DOI: 10.1515/nanoph-2021-0139