• 研究论文 • 上一篇    下一篇



  1. (西安电子科技大学 理学院,陕西 西安 710071)
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2007-10-20 发布日期:2007-10-25

Image decomposition based on sparse representations and a projected regularization method

JIANG Ling-ling;YIN Hai-qing;FENG Xiang-chu

  1. (School of Science, Xidian Univ., Xi′an 710071, China)
  • Received:1900-01-01 Revised:1900-01-01 Online:2007-10-20 Published:2007-10-25


结合稀疏表示和投影正则化方法,提出了一种将图像分解为纹理和结构部分的新方法.该方法的基本思想是用两个适当的字典:一个用来描述纹理部分——对偶树复小波变换,另一个用来描述结构部分——基于投影正则化方法的二代曲线波变换,其中投影正则化方法可以很好地指引分解过程, 减少伪吉布斯现象.这两个字典本身是互不相关的,只对它们所描述的部分得到稀疏表示,对另外一部分得不到稀疏表示.实验结果表明, 该算法即节省了运算时间, 又很好地将图像的纹理和结构分开,特别是当图像含有噪声时,它可以很好地将纹理和噪声分开.

关键词: 曲线波, 对偶树复小波变换, 全变分, 纹理, 基跟踪

Abstract: A novel method is presented for separating images into texture and cartoon parts based on sparse representations and a projected regularization scheme. The basic idea presented in this paper is the use of two appropriate dictionaries, one for the representation of texture parts-the dual tree complex wavelet transform and the other for the cartoon parts-the second generation of curvelet transform followed by a projected regularization method which is employed to better direct the separation process and reduce the pseudo-Gibbs oscillations. Both dictionaries are chosen such that they lead to sparse representations over one type of image-content and several experimental results show that the algorithm’s performance is validated.

Key words: curvelet, dual tree complex wavelet transform, total variation, texture, basis pursuit


  • TN911