Direct imaging is currently the only available method to characterize the atmosphere of long-period exoplanets (≥1 AU). This technique is difficult to carry out because it requires high contrasts at short angular separations (10^6-10^10 at ~0,2''). During my PhD thesis, I focused on two instruments : the SPICES mission and the VLT/NaCo instrument. SPICES (Spectro-Polarimetric Imaging and Characterization of Exoplanetary Systems) is a project of space coronagraph which aims to characterize in spectro-polarimetry cold exoplanets and faint circumstellar disks (≥1 zodi). I determined the science performance of the mission using numerical simulations. SPICES could characterize Jupiters at separations ≤5 AU and distances ≤10 pc, super-Earths at 1-2 AU for a few stars within 4-5 pc and possibly Earths around α Centauri A. Considering all stellar types, I estimated an amount of ~300 potential targets. In the second part of my PhD thesis, I analyzed images taken in coronagraphic and differential imaging modes of the NaCo instrument. The objective was to search for gas giant planets as cool as 500-1300 K at separations of 5-10 AU around a selected sample of young and nearby stars (<200 Myr, <25 pc). Median detection limits give contrasts of ~10−5 at 1'', corresponding to effective temperatures and masses of 1100 K and 10 MJ, and complement statistical studies based on radial velocities.
