Authors

Marie Ygouf (1,2), Laurent Mugnier (1), Jean-François Sauvage (1), Thierry Fusco (1), David Mouillet (2), Jean-Luc Beuzit (2)

Affiliations

(1) ONERA; (2) LAOG

Abstract

The next-generation instruments of 8m-class telescopes such as SPHERE (Spectro Polarimetric High contrast Exoplanet REsearch) and GPI (Gemini Planet Imager) should dramatically improve the direct detection and spectral characterization of exoplanets. One of the instruments of SPHERE will be an Integral Field Spectrograph (IFS), which will produce multispectral images made up of two spatial and one spectral dimensions. In these multispectral images, the star is not completely cancelled by the AO-corrected coronagraphic system because of residual aberrations of the latter. In particular, the star image comprises quasi-static speckles that must be disentangled from the planet signal in order to get the sought information: is there a planet, where is it and what is its spectrum? To this aim, we are developing a specific image post-processing method using a Bayesian inverse problem approach. The essential required building block of such a method is a data model (often called "direct model") with a minimal number of unknown parameters. We propose an approximate analytical direct model of a long-exposure star image for an AO-corrected coronagraphic imaging system and we present some preliminary numerical simulations to validate this approach.