计算机科学 ›› 2014, Vol. 41 ›› Issue (8): 301-305.doi: 10.11896/j.issn.1002-137X.2014.08.064

• 图形图像与模式识别 • 上一篇    下一篇

基于结构信息的RPCA图像去噪

郑秀清,何坤,张健   

  1. 四川师范大学信息技术学院 广汉618300;四川大学计算机学院 成都610065;四川师范大学网络与通信技术研究所 成都610066
  • 出版日期:2018-11-14 发布日期:2018-11-14
  • 基金资助:
    本文受四川省科技支撑计划项目(2013SF0157)资助

Image Denoising by Principal Component Analysis with Structural Information

ZHENG Xiu-qing,HE Kun and ZHANG Jian   

  • Online:2018-11-14 Published:2018-11-14

摘要: 图像在采集、存储和传输过程中不可避免地受到噪声攻击。鉴于无噪声图像对象形成的物理机理,在灰度图像中不同对象总是通过子块结构及其空间分布特性表征出来。子块结构具有某种规律或周期的像素变化,而噪声的分布表现出随机特性。为了在抑制噪声的同时保护对象结构,文中提出基于结构信息的鲁棒主成分分析(RPCA)图像去噪方法。该方法从高质量的图像集中提取结构信息样本,建立结构信息基元库,对不同结构信息基元集分别进行RPCA变换,挖掘其稀疏表示的变换核函数,用于图像去噪。实验表明,基于结构信息的RPCA图像去噪方法在去除噪声的同时,能有效保护图像的结构信息。

关键词: 结构信息,RPCA,图像去噪

Abstract: The noise is inevitable in the process of image acquisition,image storage and image transmission.To suppress the noise effectively and preserve the structural information of the image,a new method based on structural information was presented.The method extracts the sub-block sample from the pure image to build the structural element library.The structure elements are regarded as the sample of the same general.By RPCA transform,the parse representation kernel function of the structure element library of the same type samples is built adaptively which can smooth the image and preserve the edge effectively.The method analyses the distributions of noise and structure information in kernel space from the theory.Experimental results show that the proposed algorithm can suppress the noise of the image and protect the structural information of the image more effectively.

Key words: Structural information,RPCA,Image denoising

[1] Stanimirova I,Walczak B,Massart D L,et al.A three-way method for data analysis[C]∥Application to environmental data,Chem-ometrics and Intelligent Laboratory Systems.2004(73):219-233
[2] 黄宝贵,马春梅,卢振泰.新的形态学图像降噪方法[J].计算机应用,2011(3):757-759
[3] Tatsuoka K S,Tyler D E.The uniqueness of S and M-functio-nals under non-elliptical distributions[C]∥The Annals of Statistics.2000,28:1219-1243
[4] 焦峰,毕硕本,赵英男,等.基于自相似性和小波分析的图像增强与去噪[J].计算机科学,2010,37(10):242-245
[5] Hubert M,Rousseeuw P J,Verboven S.fast method for robust principal components with application to chemo metrics[C]∥Chemometrics and Intelligent Laboratory Systems,2002,60:101-111
[6] Croux C,Ruiz-Gazen A.High breakdown estimators for principal components:the projection-pursuit approach revisited[J].Journal of Multivariate Analysis,2005,7(4):206-226
[7] 焦李成.图像多尺度几何分析理论与应用[M].西安:西安电子科技大学出版社,2008
[8] Pea D,Prieto F J.Multivariate outlier detection and robust covariance estimation[C]∥Technometrics.2001,(41):286-300
[9] Olshausen B A,Field D J.Emergence of simple-cell receptivefield properties by learning a sparse code for natural images[J].Nature,1996,1(6583):1129-1159
[10] Askari E,Moghadam A M E.A fuzzy measure for objecttiveevaluation of interactive image segmentation algorithms[C]∥IEEE International Conference on Signal and Image Processing Applications (ICSIPA).2011:260-264
[11] Jia R Q,Zhao H.A fast algorithm for the total variation modelof image denoising[J].Advances in Computational Mathematics,2010,33(2):231-241
[12] Zhang Lei,Dong Wei-sheng,Zhang D.Two-stage image denoi-sing by principal component analysis with local pixel grouping[J].Pattern Recognition,2010,43(4):1531-1549

No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!