In this paper, the advancements in structured light beams recognition using speckle-based convolutional neural networks (CNNs) have been presented. Speckle fields, generated by the interference of multiple wavefronts diffracted and scattered through a diffuser, project a random distribution. The generated random distribution of phase and intensity correlates to the structured light beam of the corresponding speckle field. This unique distribution of phase and intensity offers an additional dimension for recognizing the encoded information in structured light. The CNNs are well-suited for harnessing this unique ability to recognize the speckle field by learning hidden patterns within data. One notable advantage of speckle-based recognition is their ability to identify structured light beams from a small portion of the speckle field, even in high noise environments. The diffractive nature of the speckle field enables off-axis recognition, showcasing its capability in information broadcasting employing structured light beams. This is a significant departure from direct-mode detection-based models to alignment-free speckle-based detection models, which are no longer constrained by the directionality of laser beams.
Intensity degenerate orbital angular momentum (OAM) modes are impossible to recognize by direct visual inspection even using available machine learning techniques. We are reporting speckle-learned convolutional neural network (CNN) for the recognition of intensity degenerate Laguerre–Gaussian (LGp , l) modes, intensity degenerate LG superposition modes, and intensity degenerate perfect optical vortices. The CNN is trained on the simulated one-dimensional far-field intensity speckle patterns of the corresponding intensity degenerate OAM modes. The trained CNN recognizes intensity degenerate OAM modes with an accuracy >99 % . Speckle-learned CNNs are also capable of recognizing intensity degenerate OAM modes even under the presence of high Gaussian white noise and atmospheric turbulence with an accuracy >97 % .
Machine learning has emerged as a powerful tool for physicists for building empirical models from the data. We exploit two convolutional networks, namely Alexnet and wavelet scattering network for the classification of orbital angular momentum (OAM) beams. We present a comparative study of these two methods for the classification of 16 OAM modes having radial and azimuthal phase profiles and eight OAM superposition modes with and without atmospheric turbulence effects. Instead of direct OAM intensity images, we have used the corresponding speckle intensities as an input to the model. Our study demonstrates a noise and alignment-free OAM mode classifier having maximum accuracy of >94 % and >99 % for with and without turbulence, respectively. The main advantage of this method is that the mode classification can be done by capturing a small region of the speckle intensity having a sufficient number of speckle grains. We also discuss this smallest region that needs to be captured and the optimal resolution of the detector required for mode classification.
Optical Speckles has many extraordinary applications like subwavelength focusing, aberration-free imaging, etc. which are not possible even with a highly coherent optical field. This makes it necessary to study the fundamental properties of such Optical Speckle fields. In the recent past, the polarization correlation vortex phase was experimentally realized in vector speckle field generated by scattering of Poincare beam. Higher-order correlations have also been studied in such vector speckle fields. Moving further into this direction, we have studied the first-order polarization correlations in the focused vector speckle field. We have generated a vector speckle field by scattering of Poincare beam. Which is then allowed to be focused using a spherical and a cylindrical lens. The focused vector speckle field intensities at different planes around the focal plane were recorded to get polarization correlations at each plane. It was observed that the charge of the input vortex beam before scattering is still present in the polarization correlation of the focused vector speckle field. We have also observed charge inversion of polarization correlation vortex focused through a cylindrical lens. The importance of this study relies in the fact that it provides, with supporting experimental and simulation results, that the polarization correlation obeys the wave equation. It could find application in optical image processing while analyzing any optical data, to find information about the source of the speckles, etc.
Orbital angular momentum (OAM) beams have the potential to increase the information-carrying capacity because of the extra degrees of freedom associated with them. Traditional methods for mode detection and de-multiplexing are complex and require expensive optical hardware. We propose a very simple and cost effective deep learning based model for demultiplexing OAM modes at the receiver. In this method we have used a random phase mask of known inhomogeneity to generate a scattered field of OAM mode and the intensity images of these scattered field are used as an input to the Convolutional Neural Network. The model is trained for various Laguerre-Gaussian (𝐿𝐺𝑝𝑙) modes carrying OAM with 𝑝 = 0 and 𝑙 = 1,2,3,4,5,6,7,8. The model is tested for various set of images and the overall accuracy of each dataset is <99%. To demonstrate the proof of concept we simulated an experiment to generate the speckle field at the receiver of optical communication system for demultiplexing OAM modes and decoding the 3-bit information.
The slit diffraction of circular OV beams is studied both theoretically and by experiment, with explicit involvement of the incident beam convergence or divergence (finite value of the wavefront curvature radius). Based on the example of Laguerre-Gaussian mode with zero radial index and non-zero azimuthal index m we confirm that the far-field diffraction pattern contains exactly |m| bright lobes elongated orthogonally to the slit (which was reported previously) and show that the far-field profile possesses an asymmetry with respect to the slit axis depending on the wavefront curvature (which is a new result). Being combined, these features enable simple and efficient means for the simultaneous express diagnostics of the magnitude and the sign of the OV topological charge, which can be useful in many OV applications, including the OV-assisted metrology and information processing.
Topological structure of monstar in π-symmetric fields with index Ic = +1/2 and three radial lines ending at C-point is an intermediate structure having properties of lemon and star. We experimentally realized monstar pattern in polarization ellipse orientation via three different routes, from lemon pattern using topologically-invariant squeezing and / or rotation transformations. Our results suggest that lemon and monstar can smoothly transform into each other under any or combination of these transformations leading to one interpretation that monstar is an anisotropic lemon.
We present newly conceived liquid-crystal-based retardation waveplates in which the optic axis distribution has a “superelliptically” symmetric azimuthal structure with a topological charge q superimposed. Such devices, named superelliptical q-plates, act as polarization-to-spatial modes converters that can be used to produce optical beams having peculiar spiral spectra. These spectra reflect the topological charge of the optic axis distribution as well as the symmetry properties of the underlying superellipse. The peculiar capability of q-plates of producing optical modes entangled with respect to spin and orbital angular momentum is here extended to superelliptical q-plates in order to create more complex optical modes structurally inseparable with respect to polarization and spatial degrees of freedom. Such superelliptical modes can play a crucial role in studying polarization singularities or to develop polarization metrology.
Polarization structured optical beams have half-integer topological structures: star, lemon, monstar in π-symmetric polarization ellipse orientation tensor field and integer-index topological structures: saddle, spiral, node in 2π-symmetric Poynting vector field. Topological approach to study the polarization structured optical beams is carried out and presented here in some detail. These polarization structured light beams are demonstrated to be the best platform to explore the topological interdependencies. The dependence of one type of topological structure on the other is used to control the Poynting vector density distribution and locally enhance the angular momentum density as compared to its constituent beam fields.
Interference of fiber eigen vector modes of different phase and spatial variation of polarization gives rise to
different types of polarization singularities – isolated C-point surrounded by star / lemon type polarization
morphology patterns, a dipole or two C-points of same index – in 2D polarization fields. In this context, fiber
modal decomposition refers to identifying the constituent modes, their relative amplitude and the phase
relationship among them in the fiber output. The size and location of the L-contour and the location of Cpoints
determine the relative amplitude and the orientation of the polarization morphology pattern provides
information regarding the relative phase difference between the constituent vector modes. We use these
aspects of polarization singularity to demonstrate a novel fiber modal decomposition method.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.