Diffractive optical networks unify wave optics and deep learning to all-optically compute a given machine learning or computational imaging task as the light propagates from the input to the output plane. Here, we report the design of diffractive optical networks for the classification and reconstruction of spatially overlapping, phase-encoded objects. When two different phase-only objects spatially overlap, the individual object functions are perturbed since their phase patterns are summed up. The retrieval of the underlying phase images from solely the overlapping phase distribution presents a challenging problem, the solution of which is generally not unique. We show that through a task-specific training process, passive diffractive networks composed of successive transmissive layers can all-optically and simultaneously classify two different randomly-selected, spatially overlapping phase images at the input. After trained with ~550 million unique combinations of phase-encoded handwritten digits from the MNIST dataset, our blind testing results reveal that the diffractive network achieves an accuracy of >85.8% for all-optical classification of two overlapping phase images of new handwritten digits. In addition to all-optical classification of overlapping phase objects, we also demonstrate the reconstruction of these phase images based on a shallow electronic neural network that uses the highly compressed output of the diffractive network as its input (with e.g., ~20-65 times less number of pixels) to rapidly reconstruct both of the phase images, despite their spatial overlap and related phase ambiguity. The presented phase image classification and reconstruction framework might find applications in e.g., computational imaging, microscopy and quantitative phase imaging fields.
Implementation of nearly arbitrary spatially-varying polarization transformations: a non-diffractive and non-interferometric approach using spatial light modulators
All-Optical Synthesis of an Arbitrary Linear Transformation Using Diffractive Surfaces
University of California, Los Angeles(加利福尼亚大学洛杉矶分校);California NanoSystems Institute(加州纳米系统研究所)
本文设计了一种全光学衍射表面,用于在输入(N_i)和输出(N_o)之间进行任意复值线性变换,其中N_i和N_o分别表示输入和输出视场(FOV)处的像素数。首先,我们考虑一个单一的衍射表面,并使用基于矩阵伪逆的方法来确定衍射特征/神经元的复值透射系数,以全光学地执行期望的/目标线性变换。除了这种无数据设计方法之外,我们还考虑了一种基于深度学习的设计方法,通过使用与目标变换相对应的输入/输出场的示例来优化衍射表面的透射系数。我们比较了使用无数据设计以及数据驱动(基于深度学习)衍射设计实现的全光学变换误差和衍射效率,以全光学执行(i)任意选择的复值变换,包括酉变换,非酉变换和不可逆变换,(ii)2D离散傅立叶变换,(iii)任意2D置换运算,以及(iv)高通滤波相干成像。我们的分析表明,如果空间设计的衍射特征/神经元的总数(N)为N_i x N_o或更大,则两种设计方法都成功地实现了目标变换的全光学实现,实现了可忽略的误差。然而,与无数据设计相比,发现基于深度学习的衍射设计对于给定的N可以实现更大的衍射效率,并且它们的全光学变换对于N < N_i x N_o更准确。这些结论通常适用于采用空间工程衍射表面的各种光学处理器。
Combined spatially and temporally multiplexed photonic reservoir computer with a diffractively coupled VCSEL-array
University of Strathclyde(斯特拉斯克莱德大学);Instituto de Física Interdisciplinar y Sistemas Complejos, IFISC (UIB-CSIC)(跨学科物理与复杂系统研究所,IFISC(巴利阿里群岛大学-西班牙国家研究委员会))
Polarization Multiplexed Diffractive Computing: All-Optical Implementation of a Group of Linear Transformations Through a Polarization-Encoded Diffractive Network
University of California, Los Angeles(加州大学洛杉矶分校);California NanoSystems Institute (CNSI)(加州纳米系统研究所(CNSI))
由于机器学习的变革性进展,光学计算的研究最近引起了人们的极大关注。在不同的方法中,由空间工程透射表面组成的衍射光学网络已被证明用于全光学统计推断和使用无源自由空间光学层执行任意线性变换。在这里,我们介绍了一种偏振复用衍射处理器,通过使用深度学习训练的单个衍射网络来全光学地执行多个任意选择的线性变换。在该框架中,预先选择的线性偏振器的阵列被定位在各向同性的可训练透射衍射材料之间,并且不同的目标线性变换(复值)被唯一地分配给输入/输出偏振状态的不同组合。通过使用与分配给不同输入/输出偏振组合的复值线性变换中的每一个对应的输入/输出场的数千个示例,经由深度学习和误差反向传播来训练和优化该偏振复用衍射网络的传输层。我们的结果和分析表明,当可训练的衍射特征/神经元数目(N)接近N_p x N_i x N_o时,单个衍射网络可以成功地逼近并全光学地实现一组任意选择的目标变换,误差可以忽略不计,其中N_i和N_o分别表示输入和输出视场的像素数,N_p是指分配给不同输入/输出偏振组合的唯一线性变换的数目。这种偏振复用全光衍射处理器可以在光学计算和基于偏振的机器视觉任务中找到各种应用。
Wavelength-Multiplexed 2D Beam Steering via a Passive Diffractive Network
Electrical and Computer Engineering Department, University of California, Los Angeles(加州大学洛杉矶分校电气与计算机工程系);Bioengineering Department, University of California, Los Angeles(加州大学洛杉矶分校生物医学工程系);California NanoSystems Institute (CNSI), University of California, Los Angeles(加州大学洛杉矶分校加州纳米系统研究所)
Subwavelength Imaging using a Solid-Immersion Diffractive Optical Processor
Harbin Institute of Technology (Shenzhen)(哈尔滨工业大学(深圳));University of California, Los Angeles(加州大学洛杉矶分校);California NanoSystems Institute (CNSI)(加州纳米系统研究所);Peking University(北京大学);École Polytechnique Fédérale de Lausanne (EPFL)(洛桑联邦理工学院);David Geffen School of Medicine, University of California, Los Angeles(加州大学洛杉矶分校大卫·格芬医学院)
Structural Vibration Monitoring with Diffractive Optical Processors
University of California, Los Angeles(加州大学洛杉矶分校);California NanoSystems Institute(加州纳米系统研究所);California Geological Survey(加州地质调查局);California Department of Conservation(加州自然保护部)
Diffractive lensless imaging with optimized Voronoi-Fresnel phase
King Abdullah University of Science and Technology(阿卜杜拉国王科技大学);Institute of Automation, Chinese Academy of Sciences(中国科学院自动化研究所);University of Chinese Academy of Sciences(中国科学院大学)
Dynamic Coherent Diffractive Imaging Using Only a Support Constraint in the Complex Plane
Massachusetts Institute of Technology(麻省理工学院);Fujikura Ltd.(藤仓株式会社);Stony Brook University(石溪大学);Brookhaven National Laboratory(布鲁克海文国家实验室);Northwestern University(西北大学);Singapore–MIT Alliance for Research and Technology (SMART) Centre(新加坡-麻省理工学院研究与技术创新中心)
An epsilon-near-zero-based nonlinear platform for ultrafast re-writable holography
University of Ottawa(渥太华大学);Tampere University(坦佩雷大学);University of Rochester(罗切斯特大学);Yale University(耶鲁大学);Technische Universität Darmstadt(达姆施塔特工业大学)