中文摘要：

图像去模糊一直是图像修复中的重要问题,针对经典的去模糊方法,提出一种耦合非凸lp（0≤p〈1）范数和G范数的图像去模糊方法。该方法利用lp（0≤p〈1）范数作为正则项约束,保证了图像的稀疏性要求;利用G范数作为保真项,保证在去模糊的同时有效抑制噪声并保持图像的细小特征,同时也给出新方法基于交替最小化的有效算法。实验结果表明新模型是可行的。

英文摘要：

Image deblurring is an important problem in image restoration. For the classical deblurring method, an image deblurring method based on the integration of non-convex lp（0≤p〈1） norm and G-norm is proposed. The lp（0≤p〈1） norm is taken as the regular term constraint to ensure the sparse feature of the image, and the G-norm as the fidelity item can effectively suppress the noise and keep the small feature of image while ensuring the deblurring. The new method based on the effective algorithm of alternating minimization is given. The exoerimental results show that the new modal is feasible.