[en] During the last couple of decades, studying the cosmic microwave background (CMB) temperature anisotropies has been the most powerful probe to establish our current standard cosmological model and to constrain its parameters. In spite of the impressive agreement between the model and a large variety of observations, there are still some remaining issues that polarization measurement of the CMB could help disentangle. In particular, primordial B-modes of polarization are widely considered as being a smoking gun of the hypothetical period of inflation, and detecting them is one of the most important goal in cosmology for the next few years. The measurement of CMB polarization is one of Planck's ESA mission goals. Measuring polarization is challenging though; the relative amplitudes of the signals in temperature and polarization require a strict control of systematic effects. This thesis present my work on calibration and characterization of the systematic effects for the Planck HFI instrument, as well as a study of compact sources. I propose a new method to extract compact sources photometry, based on time domain analysis instead of maps, and apply it to study polarization and time variability of astrophysical sources seen by Planck; a particular attention is dedicated to the Crab nebula. This analysis is also used to test the HFI calibration and study various systematic effects, such as beam shapes or band-pass effects, that can affect this calibration. A method to calibrate the polarization parameters of HFI detectors (polarization efficiencies and orientations in the focal plane) is also proposed and tested on simulations. (author)