Restoration of Images with Atmospheric Perturbation Simulation

We study model-based restoration of ground-to-ground images. Such images may show strong degradations due to non-linear atmospheric turbulence effects and therefore, any efficient restoration method must take into account some modelling of the turbulence. The first part of the paper is devoted to the realistic simulation of single frames and image sequences degraded by atmospheric turbulence at two different light wavelengths. The second part deals with the comparison of several restoration methods (inverse filtering, Tikhonov regularization, Laplacian regularization and Wiener filtering); a new method is also considered.