Polarimetric compressed sensing with hollow, self-assembled diffractive films

Abstract

Sensing light's polarization and wavefront direction enables surface curvature assessment, material identification, shadow differentiation, and improved image quality in turbid environments. Traditional polarization cameras utilize multiple sensor measurements per pixel and polarization-filtering optics, which result in reduced image resolution. We propose a nanophotonic pipeline that enables compressive sensing and reduces the sampling requirements with a low-refractive-index, self-assembled optical encoder. These nanostructures scatter light into lattice modes, which encode the wavefront direction and the polarization ellipticity in the linearly-polarized components of the diffracted, interference patterns. Combining optical encoders with a neural network, the system predicts pointing and polarization when the interference patterns are adequately sampled. A comparison of ``ordered'' and ``random'' optical encoders shows that the latter both blurs the interference patterns and achieves higher resolution. Our work centers on the unexpected modulation and spatial multiplexing of incident light polarization by self-assembled hollow nanocavity arrays as a class of materials distinct from traditional metasurfaces that will not only enable encoding for polarization and optical computing but also for compressed sensing and imaging.

43 pages, 17 figures

References in corpus (11)

• Spin-orbit interactions of light
• Sparsity and Incoherence in Compressive Sampling
• Meta-Gratings: Beyond the Limits of Graded Metasurfaces for Wavefront Control
• Learning to see through multimode fibers
• An optical neural network using less than 1 photon per multiplication
• Generalized Hartmann-Shack array of dielectric metalens sub-arrays for polarimetric beam profiling
• Imaging With Nature: Compressive Imaging Using a Multiply Scattering Medium
• Electromagnetic Spin-Orbit Interactions via Scattering
• Polarization Multiplexed Diffractive Computing: All-Optical Implementation of a Group of Linear Transformations Through a Polarization-Encoded Diffractive Network
• Small-brain neural networks rapidly solve inverse problems with vortex Fourier encoders
• Computational Cannula Microscopy of neurons using Neural Networks