关键词: 超分辨率, 盲重建, 模糊核估计, 罚函数, 增广拉格朗日法

Abstract: An adaptive bi-lp-l2-norm based blind super-resolution method is proposed to improve the quality of a low-resolution blurred image. The proposed blind super-resolution method splits the solution into independent blur-kernel estimation and non-blind super-resolution sub-processes. In the blur-kernel estimation sub-process, the bi-lp-l2-norm regularization is imposed on both the sharp image and the blur-kernel. Moreover, by introducing a threshold segmentation of image gradients, the lp norm and the l2 norm constraints on the sharp image are adaptively combined. With the estimated blur-kernel, the non-blind super-resolution method based on non-locally centralized sparse representation is used to construct the final high-resolution image. The experimental results demonstrate a superior performance of the proposed method in terms of kernel estimation accuracy and reconstructed image quality.

Key words: super-resolution, blind reconstruction, blur-kernel estimation, penalty function, augmented Lagrangian method