Of the 1800 exoplanets detected to date, only 50 were by direct imaging. However, by allowing the observation of circumstellar disks and planets (sometimes simultaneously around the same star, as in the case of β-pictoris), this method is a fundamental tool for the understanding of planetary formation. In addition, direct access to the light of the detected objects allows spectroscopy, paving the way for the first time to the chemical and thermal analysis of their atmosphere and surface. However, direct imaging raises specific challenges : accessing objects fainter than their star (with a ratio up to 108 to 1011), and separated only by a fraction of arc-second. To obtain these values, several techniques must be implemented. A coronagraph, used in complement with a deformable mirror and active optical aberration correction methods, produces high-contrast images, which can be further processed by differential imaging techniques. My PhD thesis work took place at the intersection of these techniques. At first, I analyzed, in simulation and experimentally on the THD –french acronym for very high contrast– bench of the Paris Observatory, the performance of the self-coherent camera, a wavefront sensing technique used to correct the optical aberrations in the focal plane. I managed to obtained high-contrast zones (called dark holes) with performance up to 3.108 between 5 and 12 λ/D, in monochromatic light. I also started an analysis of the performance in narrow spectral bands. In the second part of my thesis, I applied the latest differential imaging techniques to high contrast images from another coronagraphic instrument, NICI. The processing of these data revealed unprecedented views of the dust disk orbiting HD 15115.

Defended by the author on September 30th, 2014 at Paris Observatory for the obtention of the title of Doctor in Astronomy and Astrophysics of the Paris Diderot University. 219 pages, in french.