Super-resolution Image Reconstruction Based on Geman & McClure Norm

Super-resolution image reconstruction is an ill-posed inverse problem.In this paper we propose a new method of super resolution image reconstruction,which uses the Geman McClure norm to measure the data fidelity term,besides we propose a new priori image model based on BTV model,which can use more different directional information and choose the regularization parameters adaptively.The results indicate that the proposed method can remove noise and keep edges better,and has considerable effectiveness in terms of both visual evaluation and PSNR compared with L1 norm and L2 norm.