Direct detection of exoplanets is a very promising field in astronomy today. Planets unseen by other methods can be discovered and spectroscopically characterized. However, such detections require to overcome the high contrast (larger than 10 thousands up to several billions) and the high angular resolution (smaller than a fraction of arcsec) between the star and its companion. Coronagraphs are being used to reduce the stellar light but their efficiency is limited by wavefront aberrations that produce speckles in the final coronagraphic science plane. Classical adaptive optics usually measure aberrations in a specific estimation channel. To avoid differential aberrations in separate channels, several methods have been developed to estimate the wavefront deformations directly from the science image. Among them, the Self-Coherent Camera (SCC), which uses the principle of the star light coherence to measure the wavefront after a coronagraph. Associated with a deformable mirror, correction in closed-loop with this method has already reached high contrasts in monochromatic light. However, the widening of the wavelength bandwidth, mandatory for spectroscopic analysis of planets, limits the performance in both the estimation by the SCC and the correction by the deformable mirror. After recalling the SCC principle, we will present a theoretical analysis of this problematic and laboratory performance in polychromatic light.