J4 ›› 2010, Vol. 37 ›› Issue (5): 941-946.doi: 10.3969/j.issn.1001-2400.2010.05.029

• Original Articles • Previous Articles     Next Articles

Generalized nonlocal mean denoising research based on the wavelet domain

FENG Xiang-chu1;LIU Tao1;LI Ya-feng1,2   

  1. (1. School of Science, Xidian Univ., Xi'an  710071, China;
    2. Dept. of Computer Sci., Baoji Univ. of Arts and Sci., Baoji  721007, China)
  • Received:2009-04-10 Online:2010-10-20 Published:2010-10-11
  • Contact: FENG Xiang-chu E-mail:xcfeng@mail.xidian.edu.cn

Abstract:

The statistics of image wavelet coefficients is non-Gaussian and can be described by generalized Gaussian distribution (GGD). The paper investigates the issues of the GGD statistical model for wavelet coefficients in a subband and the corresponding parameter estimation. The estimated parameters are used to define a generalized nonlocal mean which allows us to restore the original image. A nonlocal mean denoising algorithm in the wavelet domain based on the GGD statistical model is proposed. Simulation results indicate that the proposed method outperforms the others by 1.5~3.3dB in the PSNR, and keeps a better visual result in edges information reservation as well.

Key words: wavelet coefficients, generalized Gaussian distribution, nonlocal means algorithm, image denoising