基于非局部全变差模型的图像复原方法

面向图像复原问题，本文提出一种非局部的全变差图像复原方法。该方法将传统的全变差模型拓展为非局部变差模型，充分利用非局部子块对结构的保持作用，进一步提高复原图像的质量。此外，为了解决上述的非局部全变差模型，本文引入算子运算简化目标函数, 再利用迭代的Splitting算法对其进行交替求解，提高收敛精度。实验结果表明，本文算法在视觉效果和客观评价指标两方面均优于传统算法。

Image restoration based on nonlocal total variation

In this paper, we proposed a novel nonlocal total variation method for image restoration. The proposed method extended the conventional total variation to a novel nonlocal model and made full use of the structure protection within the nonlocal patches to further improve the restoration quality. In addition, to solve the proposed nonlocal total variation model, we introduced the operator to compact the objective function, and then solved it alternatively with Splitting technique to obtain high accuracy. The experiment results shows that our proposed method outperform the conventional methods on both of the evaluation and visualization.