Robust multi-frame super-resolution with non-parametric deformations using diffusion registration

In this paper, we present a new approach of super-resolution. Since almost all super-resolution problems suffer from the motion and blur estimations, new techniques are considered to improve the registration and restoration steps. The proposed method consists of a non parametric registration based on diffusion regularisation and a total variation restoration in the reconstitution step, since super-resolution reconstruction is actually an ill-posed problem. We consider that the deformation is not parametric and differs from one image to another. We also prove the existence of a solution to the two well posed problems (registration and debluring). Simulation results show the effectiveness and robustness of our algorithm against small deformations compared to other existing methods.