English

应用科学学报

• 信号与信息处理 • 上一篇    下一篇

各向异性双树复小波包变换

石 霏
  

  1. 苏州大学 电子信息学院 江苏 苏州 215021
  • 收稿日期:2008-05-09 修回日期:2008-05-26 出版日期:2008-07-31 发布日期:2008-07-31

Anisotropic Dual-Tree Complex Wavelet Packet Transform

SHI Fei
  

  1. School of Electronic Information Engineering, Suzhou University, Suzhou 215021, China
  • Received:2008-05-09 Revised:2008-05-26 Online:2008-07-31 Published:2008-07-31

被引次数

2

摘要： 基于双树复小波变换提出了一种新的复小波包变换。该变换引入了各向异性分解方式，由此得到的小波基在各个方向上尺度可以不同，并具有更好的方向性。这种变换可用在二维或三维数据的处理上。图像和视频去噪实验结果的PSNR值与各向同性复小波变换相比有0.1～1dB的提高，去噪后图像视觉效果更好，证实了该变换方式的优越性。

关键词: 复小波变换, 各向异性分解, 去噪

Abstract:

A new type of complex wavelet packet transform based on the dual-tree complex wavelet transform is proposed. By introducing anisotropic decomposition into the transform, the associated wavelet bases may have different sizes in different dimensions, and better directivity. The transform can be used on 2D or 3D data. Experiments of denoising of image and video show that PSNR are 0.1~1dB higher than those based on isotropic complex wavelet transform. Visual quality of the denoised images is improved. This indicates advantages of the proposed decomposition method.

Key words: complex wavelet transform, anisotropic decomposition, denoising