Field Compressed Sensing

Published

January 17, 2020

Author(s)

Anthony B. Kos, Fabio C. Da Silva, Jason B. Coder, Craig W. Nelson, Grace E. Antonucci, Archita Hati

Abstract

Imaging solutions based on wave scattering seek real-time performance, high dynamic range, and spatial accuracy at scales spanning from nanometers to thousands of kilometers. Compressed sensing algorithms use sparsity to reduce sample size during image reconstruction, with optimization efforts targeting sampling pattern design and correlation properties of the image. Nevertheless, waves can sample physical domains as they propagate in space-time and are, in fact, naturally sensitive to sparsity. Field compressed sensing thus combines propagation dynamics with the inherent sensitivity of waves to sparsity to obtain information at maximum physical rates of sampling and compression. This generalizes the previous use of time of flight, projected signal strength, and fast detection, by including penetration and time correlators during reconstruction. We demonstrate image capture at frame rates of the order of 1 MHz thus opening opportunities for real-time fast imaging such as in precision space debris tracking and detonation dynamics.

Citation

Nature Communications

Pub Type

Journals

Keywords

field compressed sensing, correlation, wave scattering, detonation dynamics, image reconstruction, sparsity, physical domain sampling, high-speed image capture, space debris tracking, single-pixel camera, time of flight, time-domain Born approximation, sampling matrix, PRBS generation, through- wall imaging

Citation

Kos, A. , Da, F. , Coder, J. , Nelson, C. , Antonucci, G. and Hati, A. (2020), Field Compressed Sensing, Nature Communications (Accessed December 6, 2024)