In order to eliminate the influence on imaging system of atmospheric turbulence, a kind of Multi-frame images super-resolution restoration algorithm based on estimation of DPSF (Discrete Point Spread Function) is proposed. Firstly, the method estimated DPSF of single image by consecutive multi-frame sequence images, and then realized super-resolution restoration of degraded image by means of maximum likelihood theory and multi-frame images information. When super-resolution restoration was carried out DPSF was amended properly in order to advance convergence of the algorithm and shorten the course of restoration. The experiment shows the method can effectively realize super-resolution restoration of real short-exposure turbulence-degraded sequence images, and has strong ability in convergence, stability and resisting noise.