Computational passive imaging of thermal sources with a leaky chaotic cavity

Abstract

Computational imaging techniques are of great interest to simplify the architecture of imaging devices since random illuminations of a scene enable its reconstruction from few measurements by solving an inverse problem. Here, we present a passive system for imaging of thermal sources in the microwave range from the cross-correlation of noise signals recorded by only two channels. The channels are attached to a high Q-factor chaotic cavity with a leaky aperture on its front side. The spatial distribution of noise sources is encoded onto the broadband spectrum of the cross-correlation and can be reconstructed from the sensing matrix mapping the uncorrelated far-field speckle patterns of the cavity onto a set of frequencies. We demonstrate imaging of localized and extended thermal sources and show that the polarization of those radiations can be discriminated. Moreover, we exhibit the effectiveness of the proposed system as a compressive imaging device which exploits the natural randomness of the speckle patterns. We believe that these results are a promising step for the design of real time and low cost microwave radiometers.