Calibrations and inversion methods in high contrast imaging for the direct detection of exoplanets

The science of exoplanets is today a primary research area in astronomy. Their direct observation from the ground is nevertheless a highly delicate issue : On the first hand the flux ratio between the planet and its host star reaches 106 to 109. On the other hand, the atmospheric turbulence corrugates image formation. This kind of observations therefore requires dedicated instruments, using a large telescope, an extreme adaptive optics systems, a coronagraphic device removing star light, an optimised focal plane instrumentation (differential imager, or integral field spectrograph), but also an efficient image processing methodology. The SPHERE project gathers these different points and gives a framework to the studies done in this thesis. My thesis work consisted in developing and optimising different methods allowing an optimal detectivity. Firstly, these methods focused on the optimisation of an AO system, via the measurement and compensation of unseen aberrations. These aberrations are one of the main limitations of XAO systems. We propose simultaneously an improvement of the phase diversity approach, and a new procedure of calibration called « pseudo closed-loop ». This procedure has been validated by simulations and tested on the ONERA AO bench. An ultimate accuracy of less than 0.4nmrms per mode has been demonstrated, leading to an internal Strehl Ratio on the bench greater than 98.0% at 0.6μm. In a second time, my work consisted in developing an image processing method within the framework of differential imaging. This method, based on a Maximum A Posteriori approach, uses the multi-wavelength information from differential imager so as to simultaneously estimate the structure function of the atmosphere and the parameters of the observed object. Besides the spectral imaging, the SPHERE instrument allows the acquisition of angular differential images, using the field rotation. A method based on the theory of detection is proposed in order to efficiently process these data. In a third time, within the framework of coronagraphic imaging, I proposed a novel analytic model of a long exposure coronographic image. This model accounts for a perfect coronagraph, for residual turbulence after AO correction and for static aberrations upstream and downstream the focal mask. These methods should be implemented on instrument SPHERE, and should contribute to the discovery of new exoplanets by 2011.