Kernel-phases for high-contrast detection beyond the resolution limit

TL;DR

This paper introduces Kernel-phases, a new method for high-contrast companion detection in direct images that surpasses the classical resolution limit by leveraging phase-noise immune observables derived from pupil geometry.

Contribution

It presents Kernel-phases as a novel observable derived from pupil geometry, enabling detection of faint companions at or beyond the diffraction limit in direct imaging.

Findings

Kernel-phases can be extracted from direct images in the high-Strehl regime.

Re-analysis of archival HST/NICMOS data demonstrates successful detection of companions.

Kernel-phases outperform traditional methods in resolution and contrast detection.

Abstract

The detection of high contrast companions at small angular separation appears feasible in conventional direct images using the self-calibration properties of interferometric observable quantities. In the high-Strehl regime, available from space borne observatories and using AO in the mid-infrared, quantities comparable to the closure-phase that are used with great success in non-redundant masking inteferometry, can be extracted from direct images, even taken with a redundant aperture. These new phase-noise immune observable quantities, called Kernel-phases, are determined a-priori from the knowledge of the geometry of the pupil only. Re-analysis of HST/NICMOS archive and other ground based AO images, using this new Kernel-phase algorithm, demonstrates the power of the method, and its ability to detect companions at the resolution limit and beyond.