Fast super-resolution on moving objects in video sequences

We present a super-resolution (SR) color freeze frame of small moving objects in video sequences. In the last two decades, all SR methods except one focus on the case of rigid scene. We propose a fast and robust method that performs SR reconstruction on tracked objects. After an affine registration of the regions of interest of objects, a non-uniform interpolation in a high-resolution grid is performed, then, a restoration. Experiments on real data validate our choices, and confirm the robustness and the rapidity of our method.