Super-resolution Reconstruction Based on Irregular Sample and Local Statistics

A novel super-resolution reconstruction algorithm based on irregular sample and local statistics is proposed in this paper. The pixels of the low resolution images are matched on the high resolution grid of the reference image which results in the irregular sample of the reconstructed high resolution image. Then Benedetto-Heller theorem is used for the reconstruction in which a new weighted reconstruction filter considering the image local statistics is proposed. Experimental results demonstrate the proposed method is better than that of the conventional algorithms in terms of visual inspection and convergence speed. It has an improvement of 1-2 dB in PSNR over other approaches.