Deep Learning Enabled Real Time Speckle Recognition and Hyperspectral Imaging using a Multimode Fiber Array

Abstract

We demonstrate the use of deep learning for fast spectral deconstruction of speckle patterns. The artificial neural network can be effectively trained using numerically constructed multispectral datasets taken from a measured spectral transmission matrix. Optimized neural networks trained on these datasets achieve reliable reconstruction of both discrete and continuous spectra from a monochromatic camera image. Deep learning is compared to analytical inversion methods as well as to a compressive sensing algorithm and shows favourable characteristics both in the oversampling and in the sparse undersampling (compressive) regimes. The deep learning approach offers significant advantages in robustness to drift or noise and in reconstruction speed. In a proof-of-principle demonstrator we achieve real time recovery of hyperspectral information using a multi-core, multi-mode fiber array as a random scattering medium.

12 pages, 6 figures + Appendix of 5 pages and 5 figures

References in corpus (17)

• Light fields in complex media: mesoscopic scattering meets wave control
• Deep speckle correlation: a deep learning approach towards scalable imaging through scattering media
• Learning to see through multimode fibers
• Imaging With Nature: Compressive Imaging Using a Multiply Scattering Medium
• Quantum-Secure Authentication with a Classical Key
• Single-shot multispectral imaging with a monochromatic camera
• Using a multimode fiber as a high resolution, low loss spectrometer
• High-resolution and broadband all-fiber spectrometers
• Pushing the limits of optical information storage using deep learning
• Two-photon imaging through a multimode fiber
• Broadband multimode fiber spectrometer
• Computational multi-spectral video imaging
• Programmable multiport optical circuits in opaque scattering materials
• An ultrafast reconfigurable nanophotonic switch using wavefront shaping of light in a nonlinear nanomaterial
• Single-photon entanglement generation by wavefront shaping in a multiple-scattering medium
• Speckle-based hyperspectral imaging combining multiple scattering and compressive sensing in nanowire mats
• Snapshot fiber spectral imaging using speckle correlations and compressive sensing

Cited by in corpus (12)

• Deep learning in nano-photonics: inverse design and beyond
• Controlling light propagation in multimode fibers for imaging, spectroscopy and beyond
• Deep learning enabled laser speckle wavemeter with a high dynamic range
• Deep learning enabled strategies for modelling of complex aperiodic plasmonic metasurfaces of arbitrary size
• Femtometer-resolved simultaneous measurement of multiple laser wavelengths in a speckle wavemeter
• Deep Learning Cell Imaging through Anderson Localizing Optical Fibre
• Neural-network-assisted in situ processing monitoring by speckle pattern observation
• Monitoring MBE substrate deoxidation via RHEED image-sequence analysis by deep learning
• Determining intrinsic sensitivity and the role of multiple scattering in speckle metrology
• Polarized hyperspectral imaging with single fiber bundle via incoherent light transmission matrix approach
• Assessing the properties of a colloidal suspension with the aid of deep learning
• High-Fidelity and High-Speed Wavefront Shaping by Leveraging Complex Media