Computational ghost imaging for remote sensing

Abstract

Computational ghost imaging is a structured-illumination active imager coupled with a single-pixel detector that has potential applications in remote sensing. Here we report on an architecture that acquires the two-dimensional spatial Fourier transform of the target object (which can be inverted to obtain a conventional image). We determine its image signature, resolution, and signal-to-noise ratio in the presence of practical constraints such as atmospheric turbulence, background radiation, and photodetector noise. We consider a bistatic imaging geometry and quantify the resolution impact of nonuniform Kolmogorov-spectrum turbulence along the propagation paths. We show that, in some cases, short-exposure intensity averaging can mitigate atmospheric-turbulence-induced resolution loss. Our analysis reveals some key performance differences between computational ghost imaging and conventional active imaging, and identifies scenarios in which theory predicts that the former will perform better than the latter.