Beam profiler network (BPNet): a deep learning approach to mode demultiplexing of Laguerre–Gaussian optical beams

Abstract

The transverse field profile of light is being recognized as a resource for classical and quantum communications for which reliable methods of sorting or demultiplexing spatial optical modes are required. Here, we demonstrate, experimentally, state-of-the-art mode demultiplexing of Laguerre-Gaussian beams according to both their orbital angular momentum and radial topological numbers using a flow of two concatenated deep neural networks. The first network serves as a transfer function from experimentally-generated to ideal numerically-generated data, while using a unique "Histogram Weighted Loss" function that solves the problem of images with limited significant information. The second network acts as a spatial-modes classifier. Our method uses only the intensity profile of modes or their superposition, making the phase information redundant.

Transformation and Phase Retrieval of Electromagnetic Fields between a Plane and an Arbitrary Surface Using Machine Learning

“…As such, we could expect an improvement in the system performance if the network was trained on experimental data, instead of on numerically generated data with the obvious disadvantage that the network would be tailored for a specific optical setup, including its aberrations and misalignment. 6 Phase Retrieval Results. Next, we demonstrate phase retrieval of the field at the input surface through knowledge of the amplitude (intensity) at the arbitrary target surface.…”

“…The difference between the numerical and experimental results can be attributed to aberrations in the optical setup. As such, we could expect an improvement in the system performance if the network was trained on experimental data, instead of on numerically generated data with the obvious disadvantage that the network would be tailored for a specific optical setup, including its aberrations and misalignment …”

“…In recent years, deep-learning (DL) networks have become an integral tool for solving complex problems in many fields of optics, such as the design of meta-materials, − spatial-mode demultiplexing, − super-resolution microscopy, , image retrieval from speckle patterns, , and phase retrieval, to name a few. One of the main contributions of DL networks is their ability to solve inverse problems, which usually do not possess an analytical solution.…”

Speckle-learned convolutional neural network for the recognition of intensity degenerate orbital angular momentum modes

“…To overcome this problem machine learning and deep learning methods are been proposed and demonstrated for rapid detection of OAM modes with better accuracy 14 – 21 Despite better accuracy, these are alignment limited and one needs to capture the entire mode. These limitations were overcome in the recent demonstration on the speckle-based CNN 22 – 24 and wavelet scattering network 25 …”

Dual-Output Mode Analysis of Multimode Laguerre-Gaussian Beams via Deep Learning

“…However, coherence measurements can only determine the mode amplitude and cannot obtain the mode phase information; intensity recording methods take more time and are difficult to achieve in real-time. In recent years, the convolutional neural network (CNN) has been widely used in fields related to optical imaging [17][18][19][20], and likewise in mode recognition [21,22], demultiplexing [23][24][25] Optics 2021, 2 88 of OAM beams. Even in propagation environments, such as atmospheric turbulence and underwater, CNNs have shown good accuracy performance [26][27][28][29].…”