Super-resolution Reconstruction Base on a New Image Registration

The registration error is one of the important factors to the image quality reconstructed by super-resolution, because over sampling and other factors always cause the actual low-resolution images are severely aliased, the reconstructed images obtain a lower reconstruction precision. For this case, the method combine registration and reconstruction can overcome registration errors caused by aliasing. This paper uses the more efficient variable projection procedure instead the cyclic coordinate descent optimization procedure based on the regularization equation proposed by Lee. The experimental results show the effectiveness of this algorithm.