一种基于L^p范数的自适应图像去噪模型  

An adaptive Lp-based image denoising model

作　　者：王际朝[1] 李维国[1] 

机构地区：[1]中国石油大学数学与计算科学学院,山东东营257061

出　　处：《中国石油大学学报（自然科学版）》2008年第2期155-158,共4页Journal of China University of Petroleum(Edition of Natural Science)

摘　　要：为了克服L2范数的去噪模型不能保持图像边缘和L1范数的总变差(total variation,TV)模型容易引起梯子现象的缺陷,提出了一个基于Lp(1≤p≤2)范数的自适应模型,并且给出了自适应函数p(x,y)的构造公式。将Primal-Dual方法应用到自适应模型的求解中,给出了求解自适应模型的公式。数值试验结果表明,自适应模型在去除噪声的同时保持了边缘的信息,且能削弱梯子现象,所恢复的图像在信噪比等方面比原来的两种模型有较大的改善。The L^2-based denoising model could not keep the edge of image. The L^1 -based total variation （TV） denoising model could produce the staircase effect. In order to combine the advantages of them, an adaptive image denoising model based on L^p （1 ≤p≤2）norm was proposed. And the adaptive function p（x,y） was given. The Primal-Dual method was used to solve the adaptive model. The numerical experimental results show that the adaptive model can remove the noise while preserving details of edge, and reduce staircase. Compared with other two models, the signal-to-noise ration of the image restored by the adaptive model improves much.

关 键 词：图像去噪 自适应模型 L^p范数 总变差 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]