变正则参数方法在带噪图像保边缘恢复中的应用

Adaptive Regularization Parameter Approach for Noisy Image Restoration with Edge Preserving

摘要: 提出变正则参数的变分方法.该方法通过选取随梯度变化的自适应的正则参数,达到去噪和保持边缘的目的.通过5点格式构造出变分模型的离散解法,可以避免格林函数的繁琐计算.

Abstract: This paper presents a variational approach using adaptive regularization parameter for keeping the information of edges and denoising largely.Through constructing a 5-points scheme of the partial differential equation （PDE）,a numerical solution can be designed to avoid the complicated computing of Green function.