In this communication, we study the statistical properties of the light intensity in direct and coronagraphic images, in the context of ground-based Extreme Adaptive Optics observations. The same approach can also be used for space observations with different scales. We show that a coronagraph only affects the perfect part of the wave and leaves the uncorrected part of the wavefront almost unaffected. This statistical model can explain the ‘speckle pinning’ effect (presence of speckles at the position of the diffraction rings), as an amplification of the speckle noise. This statistical approach can be verified on real adaptive optics data.