一种抗混叠移不变的Contourlet变换
PDF下载 (613)黄小霞,符冉迪,石大维,范亚会.一种抗混叠移不变的Contourlet变换[J].宁波大学学报(理工版),2012,25(04):29-34.DOI:
HUANG Xiao-xia,FU Ran-di,SHI Da-wei,FAN Ya-hui.A Contourlet Transform with Anti-aliasing and Shift-invariance[J].Journal of Ningbo University(Natural Science & Engineering Edition),2012,25(04):29-34.DOI:
| Title: | A Contourlet Transform with Anti-aliasing and Shift-invariance |
| 作者: | 黄小霞, 符冉迪, 石大维, 范亚会 |
| Author(s): | HUANG Xiao-xia, FU Ran-di, SHI Da-wei, FAN Ya-hui |
| 关键词: | Contourlet变换; 稀疏表示; 频谱混叠; 平移不变性 |
| Keywords: | contourlet transform; sparse representation; spectrum aliasing; shift-invariance |
| 分类号: | TN919.8 |
| 文献标识码: | A |
| 摘要: | 基于传统Contourlet变换及已有相关改进方法存在频谱混叠或移变性的缺点, 其处理后的图像会出现伪吉布斯现象, 严重影响了图像的质量, 因此提出一种既抗混叠又有平移不变性的Contourlet变换新方法来克服这两个缺点. 新方法是一种结合了抗混叠塔式滤波器组和非下采样方向滤波器组的方法. 通过比较新方法和相关Contourlet变换方法的非线性逼近性能和降噪效果, 表明新方法更能稀疏地表示图像, 且在降噪性能上有较大的提高; 它不仅有效地克服了因频谱混叠和移变性所带来的“刮痕”现象, 而且能更好地保护图像的边缘和纹理细节信息. |
| Abstract: | The traditional contourlet transform and its improved methods have been identified with some drawbacks such as spectral aliasing or shift-invariance. The processed image by these methods will appear pseudo-Gibbs phenomena, which pose adverse impacts on the quality of the image. In this paper, a modified method is proposed to overcome these two major drawbacks, which adopts contourlet transform with anti-aliasing and shift-invariance. The new method is constructed as a combination of the anti-aliasing pyramid filter bank and the nonsubsampled directional filter bank. Through comparing the nonlinear approximation performance and the denoising effect with the related contourlet transform methods, it is found that the proposed method has better sparse performance and greatly improved denoising effect. In addition, it efficiently overcomes the “scratches” phenomena caused by spectrum aliasing and shift-invariance, and has better ability to protect image’s edges and texture details. |
| 参考文献 /References: | [1] 焦李成, 谭山. 图像的多尺度几何分析: 回顾和展望[J]. 电子学报, 2003, 31(12A):1975-1981. [2] 段贵多, 李建平, 黄添喜. 图像的多尺度几何分析概述[J]. 计算机应用研究, 2007, 24(10):9-14. [3] 冯鹏, 魏彪, 米德伶. 抗混叠塔型变换的构造[J]. 电子学报, 2009, 37(11):2510-2514. [4] 易文娟, 郁梅, 蒋刚毅. Contourlet一种有效的方向多尺度分析方法[J]. 计算机应用研究, 2006(9):18-22. [5] Do M N, Vetterli M. The contourlet transform: An efficient directional multiesolution image representation [J]. IEEE Trans Image Proc, 2005, 14(12):2091-2106. [6] Lu Y, Do M N. A new contourlet transform with frequency localization[J]. IEEE Trans Image Proc, 2002, 11(10):1629-1932. [7] Cunha A L, Zhou J, Do M N. The nonsubsample Contourlet transform: Theory, design and application[J]. IEEE Trans Image Proc, 2006, 18(10):2269-2278. [8] Nguyen T T, Oraintara S. The shiftable complex direc- tionnal pyramid-part I: Theoretical Aspects[J]. IEEE Trans, 2008, 56(10):4651-4660. [9] 陈建军, 田逢春, 李灿. 基于NSCT阈值和自适应扩散的图像降噪[J]. 计算机工程, 2011, 37(12): 204-205. [10] 王军. 一种改进的小波阈值高斯噪声图像降噪方法[J]. 计算机仿真, 2011, 28(2):295-299. |
| 备注/Memo: | 收稿日期: 2012-06-01. 基金项目: 浙江省自然科学基金（Y1111061）. 第一作者: 黄小霞（1986－）, 女, 湖南永州人, 在读硕士研究生, 主要研究方向: 数字图像处理. E-mail: huangxiaoxia0000@sina.com *通讯作者: 符冉迪（1971－）, 男, 浙江慈溪人, 副教授, 主要研究方向: 图像处理和模式识别. E-mail: furandi@nbu.edu.cn 宁波大学学报（理工版）网址：http://journallg.nbu.edu.cn/ |
