中文摘要：

针对数字图像由于失焦模糊和脉冲噪音所带来的降质问题,提出了一种全新的图像复原方法.该方法将Tikhonov正则项引入到L0极小化模型,并借助交替方向法和变量分离法,给出了模型局部极小值点的快速数值求解算法.为了说明新方法的有效性,对后来的数值进行了实验,通过结合不同降质条件对新方法与其他已有方法的复原效果进行了比较.实验结果表明,新方法不仅在常规降质条件下提供了较好的复原效果,而且在面对高强度的脉冲噪音降质时,其他方法已经失效,新方法仍能提供稳健的复原结果.

英文摘要：

This paper studies the problem of image restoration of blurred image corrupted by impulse noise. We propose a novel L0 minimization restoration method combined with Tikhonov regularization and solve it by utilizing alternating direction method and variable-spliting method. With a derived iteratively upgrading algorithm,a local optimal solution can be obtained. To illustrate effectiveness of the new method, plenty of numerical experiments are presented, which shows that the proposed method is not only superior to the existing models in PSNR performance, but it can be still efficient even under high intensity impulse noise while others fail.