One crucial aspect for the science observations assisted by Adaptive Optics (AO) is the knowledge of the Point Spread Function (PSF). The PSF delivered by AO systems has a complex shape, combining spatial, spectral and temporal variability, such that its characterization is often a major limitation when analyzing AO data. The absence of reference calibrators is also common in cases like extended objects and very crowded regions. This paper presents a post-processing method (called AMIRAL) derived from blind deconvolution, which allows us to estimate the AO-PSF directly from scientific observations. AMIRAL uses an analytical PSF model (PSFAO19) and simplifies the estimation down to a few parameters. The resultant PSF is used to perform deconvolution. We first evaluate the performance of AMIRAL for PSFs retrieval with simulated data in different parameters. Then, we present a new feature by introducing a Fourier-based object model. Taking advantage of having a more realistic representation of the object, we improve both the performance and robustness of the PSF estimation, and the consequent deconvolution process. This performance gain is eventually illustrated with real observations of the asteroid Kleopatra acquired by VLT-SPHERE.
|