PhD defense of Alexis Lau

The defense will take place on Friday December 15th at 2:30pm in the amphitheater at LAM.

Zoom link: https://univ-amu-fr.zoom.us/j/82643195786?pwd=dnlZZ1dadmgreTJ4eitZNlZBNzhnUT09

Title: “Development of adaptive optics postprocessing methods: robust deconvolution of images and hyperspectral data cubes”

Abstract:
A significant challenge in ground-based astronomical observations at visible wavelengths is mitigating the impact of atmospheric turbulence. In the absence of corrective measures, even the largest terrestrial telescopes are limited to an angular resolution equivalent to a 10-20 cm telescope, translating to a resolution no finer than 0.5-1 arcsecond. Adaptive Optics (AO) technology has partially addressed this by correcting atmospheric distortions, thereby improving angular resolution towards the theoretical diffraction limit. Nonetheless, a critical limi- tation persists in obtaining an accurate Adaptive Optics Point Spread Function (AO-PSF) for effective data processing, especially when deconvolution methods are employed. Given the complex and time-varying nature of the PSF, calibration stars observed before or after the scientific observations often provide inadequate reference PSFs, leading to artefacts in the deconvolved image. Additionally, there is an emerging necessity to account for the wavelength-dependent properties of the PSF, particularly when using hyperspectral imaging systems like HARMONI. This thesis aims to enhance existing marginal PSF estimation methods by improving object priors and introducing wavelength variability into the PSF model. The refined method is more robust for observations involving resolved extended objects and is flexible to incorporate the PSF’s spectral diversity in hyperspectral imaging. Simulations employing realistic system configurations, such as VLT-like telescope and ELT-HARMONI, have been carried out to assess the characteristics and performance of this approach. Furthermore, observations from SPHERE-ZIMPOL (a visible imager that works with different wavebands) and MUSE-NFM (a hyperspectral imager) have been used to validate the method- ology within the context of realistic AO-assisted observations.

Jury :
Jessica Lu (Rapporteure), UC Berkeley
Niranjan Thatte (Rapporteur), Oxford University
Joel Vernet (Examinateur), ESO
Magali Deleuil (Présidente du jury), Aix Marseille Université
Benoit Neichel (Directeur de thèse), CNRS
Thierry Fusco (Co-Directeur de thèse), ONERA
Romain Fétick (Membre invité), ONERA