Small-brain neural networks rapidly solve inverse problems with vortex Fourier encoders

Abstract

We introduce a vortex phase transform with a lenslet-array to accompany shallow, dense, ``small-brain'' neural networks for high-speed and low-light imaging. Our single-shot ptychographic approach exploits the coherent diffraction, compact representation, and edge enhancement of Fourier-tranformed spiral-phase gradients. With vortex spatial encoding, a small brain is trained to deconvolve images at rates 5-20 times faster than those achieved with random encoding schemes, where greater advantages are gained in the presence of noise. Once trained, the small brain reconstructs an object from intensity-only data, solving an inverse mapping without performing iterations on each image and without deep-learning schemes. With this hybrid, optical-digital, vortex Fourier encoded, small-brain scheme, we reconstruct MNIST Fashion objects illuminated with low-light flux (5 nJ/cm${}^2$) at a rate of several thousand frames per second on a 15 W central processing unit, two orders of magnitude faster than convolutional neural networks.

References in corpus (4)

• Metasurface interferometry towards quantum sensors
• Probing 10 μK stability and residual drifts in the cross-polarized dual-mode stabilization of single-crystal ultrahigh-Q optical resonators
• Noninvasive nonlinear imaging through strongly-scattering turbid layers
• Spectral Representations for Convolutional Neural Networks

Cited by in corpus (6)

• Nonlinear optical encoding enabled by recurrent linear scattering
• Ultrafast Microscopy of a Plasmonic Spin Skyrmion
• Robust Position Sensing with Wave Fingerprints in Dynamic Complex Environments
• Polarimetric compressed sensing with hollow, self-assembled diffractive films
• Learning Illumination Patterns for Coded Diffraction Phase Retrieval
• Compute-first optical detection for noise-resilient visual perception