A regularization method is proposed to implement the post-processing task from the view of image restoration. The mathematical model of regularization is analyzed

also the properties of the regularization function with edge preserving. Then

the potential function is chosen and modified according to the tradeoff among the complexity

robustness and the granularity control

and the half-quadratic regularization strategy is applied for the reason of achieving the fast minimization of the energy function. Finally

the specific alternative iteration for the minimization process is described. The proposed method is adaptive to the edge details of images and can reach the minimum values fastly. The experimental results show that the method can improve the subject and visual quality of image. Compared with the MAP based algorithm

it may preserve more image details.