计算机工程 ›› 2012, Vol. 38 ›› Issue (5): 213-215.doi: 10.3969/j.issn.1000-3428.2012.05.066

• 图形图像处理 • 上一篇    下一篇

基于四元数小波变换HMT模型的图像去噪

殷 明a,b,刘 卫b   

  1. (合肥工业大学 a. 计算机与信息学院;b. 数学学院,合肥 230009)
  • 收稿日期:2011-08-12 出版日期:2012-03-05 发布日期:2012-03-05
  • 作者简介:殷 明(1962-),男,副教授、博士研究生,主研方向:小波分析,图像处理;刘 卫,硕士研究生
  • 基金项目:
    安徽省自然科学基金资助项目(11040606M06);安徽省 教育厅重点科研基金资助项目(J2010A282)

Image Denoising Based on HMT Model of Quaternion Wavelet Transform

YIN Ming   a,b, LIU Wei  b   

  1. (a. School of Computer and Information; b. School of Mathematics, Hefei University of Technology, Hefei 230009, China)
  • Received:2011-08-12 Online:2012-03-05 Published:2012-03-05

摘要: 实小波变换域中的图像去噪会产生伪Gibbs现象,为此,提出一种基于四元数小波变换的隐马尔可夫树模型(Q-HMT),并应用于图像去噪。模型结合四元数小波变换理论与HMT模型,体现小波系数层间的相关性,捕获小波系数邻域的统计特征。实验结果表明,该模型的图像去噪效果在峰值信噪比以及视觉效果上均优于经典的去噪方法。

关键词: 四元数小波变换, 实小波变, 隐马尔可夫树模型, 图像去噪

Abstract: According to the Gibbs effect of image denoising in real wavelet transform domain, this paper proposes a Hidden Markov Tree(HMT) model based on Quaternion Wavelet Transform(QWT). The QWT theory and HMT are combined in this model which effective reflect the correlation between wavelet coefficient and capture the neighborhood statistical characteristics of wavelet coefficients. Experimental results show that the model both in Peak Signal to Noise Ratio(PSNR) and in the visual effect are better than many classic denoising method.

Key words: Quaternion Wavelet Transform(QWT), real wavelet transform, Hidden Markov Tree(HMT) model, image denoising

中图分类号: