High-fidelity tomographic speckle imaging of scenes with spatially-varying point spread functions

Abstract

We describe how a two-channel speckle imaging system can be effectively employed for high-fidelity imaging of objects observed under conditions where the point spread function varies spatially across the image. We show that this variation is present in images with fields-of-view as small as an isoplanatic patch, and that ignoring it during image restoration leads to decreased fidelity in both the morphological details and the contrast of the restored image. We show that such degradation can result in false inferences on the target under study for two types of object: a high-contrast binary star system and a solar system planet (an extended object). We then demonstrate that our algorithms enable atmospheric tomography under D/r0 = 20 conditions, assuming that the atmosphere behaves as a small number of phase layers under Taylor’s frozen flow hypothesis. The layer heights are estimated via a grid search over the cost function. Height sensitivity using this method is shown to be superior to that provided by scintillation. The spatially-varying point spread functions are recovered accurately enough to enable photometry of faint companions within the isoplanatic angle and across the full ∼ 40” anisoplanatic Saturn.