Image Recovery for Low Earth Orbit by Leveraging Turbulence and Light Curves

This paper shows a novel method to characterize human-made objects in low Earth orbit (LEO) using compressed sensing on light curve measurements. The proposed approach minimizes total variation to recover a resolved object image from a fully unresolved light curve and a so-called point spread function (PSF) map. The light curves are generated through numerical wave propagation, which considers atmospheric turbulence under anisoplanatic conditions. Subsequently, the light curve model is transformed into a linear measurement model to apply compressed sensing techniques. Notably, the sensing matrix is found to be a superposition of spatially variable PSFs, which significantly downsamples the ideal object image. The proposed approach robustly recovers clear images of objects in LEO, even with imperfect PSF map estimates and Poisson noise in the light curve measurement.