Joint blur identification and high-resolution image estimation based on weighted mixed-norm with outlier rejection

We address problems of conventional super-resolution (SR) methods having the following limitations. First, most of the existing SR algorithms can not cope with local motions and hence not suitable for video sequences. Second, the blurring operator is assumed to be known in advance and constant for all the low-resolution (LR) images. Finally, SR noise is assumed to be either Gaussian or Laplacian. To solve these problems, we propose a general cost function that consists of weighted L₁- and Z₂-norms considering the SR noise model where the weights are generated from the error of registration and penalize parts that are inaccurately registered. Both the super- resolved images and blurring operators are jointly estimated. The objective and subjective results are shown to demonstrate the effectiveness of the proposed algorithm.