Video-to-video dynamic super-resolution for grayscale and color sequences

Abstract

We address the dynamic super-resolution (SR) problem ofreconstructing a high-quality set of monochromatic or colorsuper-resolved images from low-quality monochromatic, color, ormosaiced frames. Our approach includes a joint method forsimultaneous SR, deblurring, and demosaicing, this way taking intoaccount practical color measurements encountered in videosequences. For the case of translational motion and commonspace-invariant blur, the proposed method is based on a very fastand memory efficient approximation of the Kalman filter (KF).Experimental results on both simulated and real data are supplied,demonstrating the presented algorithms, and their strength.