Adaptive regularization with Lorentzian norm for image superresolution

Super-resolution reconstruction (SRR) is an effective approach for improving spatial resolution of image, which does not need change the original imaging system hardware. By introducing appropriate regularization term in the image SRR, the edge of the reconstructed image can be preserved while noise amplification being restrained. In addition, the choice of error term is important for SRR. However, how to construct a suitable cost function and regularization parameter had been an open question. In this paper, we propose an improved SRR method with Lorentzian norm combining adaptive regularization. The adaption of Lorentzian norm can resolve outlier problem and preserve edge of image. Furthermore, by adaptively selecting regularization parameter in the proposed method can avoid the randomness of trial and error method. Experimental results are on standard images show that the proposed method is effective.