基于局部邻域信息的Contourlet变换域图像去噪

Image Denoising Based on Local Information in the Contourlet Variable Domain

分析了Contourlet变换系数间的统计特性,提出了一种新的基于Contourlet变换的图像去噪算法。这种算法考虑相邻Contourlet系数间的相关性,认为某一位置"干净"的Contourlet系数不仅与这一位置本身的含噪系数有关,也与其周围邻域的含噪系数有关。实验结果表明,所提出的算法不仅能得到较满意的"线形结构",而且也能获得比现有方法更高的信噪比。

We analyze the statistical property of Contourlet coefficients and propose a new image denoising algorithm. The newalgorithm takes into account the dependent property of Contourlet coefficients, and assumes that a noise-free coefficient is not only related to the corresponding noise coefficient but also those in its local neighborhood. Experiments demonstrate the new denoising algorithm can recover the line-type structure well and produce higher PSNR than the existing algorithms.