基于稀疏性的图像去噪综述  

Overview on sparse image denoising

作　　者：郭德全[1,2] 杨红雨[1,3] 刘东权[3] 何文森[3] 

机构地区：[1]四川大学视觉合成图形图像技术国防重点实验室,成都610065 [2]内江职业技术学院机械系,四川内江641100 [3]四川大学计算机学院,成都610064

出　　处：《计算机应用研究》2012年第2期406-413,共8页Application Research of Computers

基　　金：四川省科技创新工程资助项目(2010-026);国家"973"计划资助项目(2009CB320803)

摘　　要：利用图像的稀疏与冗余表达模型去噪是当前较为新颖的去噪方法,在对国内外稀疏模型去噪文献进行理解和分析的基础上,回顾稀疏性去噪研究的发展,阐明稀疏去噪的原理与降噪模型。总结用于稀疏去噪中的各类方法,介绍利用稀疏性在图像去噪中的分解与重构过程,并将小波法去噪、多尺度几何分析法去噪、独立成分法去噪中所涉及的传统稀疏性与当前的稀疏与冗余表达模型去噪对比分析。最后基于对稀疏性去噪方法的分析,提出对稀疏去噪研究方法的一些展望。Image denoising through sparse and redundant representation modeling has been well acknowledged as an important approach of image denoising in recent years.This paper attempted to make an overview of sparse model denoising based on understanding and analysis of recent domestic and abroad literatures.To begin with,this paper reviewed the development of sparse denoising research,and clarified the principle and noise model of sparse denoising.Next,summarized several methods in procedure of sparse denoising,and introduced sparse decomposition and reconstruction in the process of image denoising.In addition,described the other denoising methods,such as the wavelet denoising method,multi-scale geometric analysis（MGA） denoising approach,independent component analysis denoising technique,and then compared and analyzd the relationships between the recent sparse and redundant representation modeling denoising method and other traditional sparse methods.Finally,pointed the problem and some future directions of sparse denoising method.

关 键 词：稀疏去噪 降噪模型 小波方法 多尺度几何分析 独立成分分量 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]