Deep Learning Enabled Transmission of Full‐Stokes Polarization Images Through Complex Media
Open Access
- 26 June 2024
- journal article
- research article
- Published by Wiley in Laser & Photonics Reviews
- Vol. 18 (11)
Abstract
Polarization images offer crucial functionalities across multiple scientific domains, providing access to physical information beyond conventional measures such as intensity, phase, and spectrum of light. However, the challenge of transmitting polarization images through complex media has restricted their application in optical communication and imaging. Here, a novel approach utilizing deep learning for the transmission of full‐Stokes polarization images through scattering media is presented. It is demonstrated that any input polarization image can be reconstructed in a single shot by employing only an intensity sensor. By supervised training of a deep neural network, high‐accuracy full‐Stokes reconstruction is achieved from the speckle pattern detected by an intensity camera. Leveraging the deep learning based polarization decoder, a polarization‐colored encoding scheme is devised to enable increased‐capacity data transmission through disordered channels. Fast, wavelength‐independent, on‐chip, polarization imaging in complex media enables the utilization of polarization‐structured light in multimode fibres and opaque materials, unlocking new possibilities in optical communication, cryptography, and quantum technology.Keywords
Funding Information
- Ministero dell’Istruzione, dell’Università e della Ricerca (PRIN n. 2022N738SA)
This publication has 53 references indexed in Scilit:
- Universal Polarization Transformations: Spatial Programming of Polarization Scattering Matrices Using a Deep Learning‐Designed Diffractive Polarization TransformerAdvanced Materials, 2023
- A Robust Basis for Multi‐Bit Optical Communication with Vectorial LightLaser & Photonics Reviews, 2023
- Polarisation optics for biomedical and clinical applications: a reviewLight: Science & Applications, 2021
- Image Transmission Through a Dynamically Perturbed Multimode Fiber by Deep LearningLaser & Photonics Reviews, 2021
- Multimode optical fiber transmission with a deep learning networkLight: Science & Applications, 2018
- Full-Stokes Imaging Polarimetry Using Dielectric MetasurfacesACS Photonics, 2018
- High-resolution grayscale image hidden in a laser beamLight: Science & Applications, 2017
- Characterizing quantum channels with non-separable states of classical lightNature Physics, 2017
- Complete experimental toolbox for alignment-free quantum communicationNature Communications, 2012
- Polarization vision: a new sensory approach to image understandingImage and Vision Computing, 1997