一种无抽样Contourlet变换的图像去噪方法

提出一种基于Contourlet变换的图像去噪方法，Contourlet变换采用无抽样形式，系数萎缩处理阈值门限与尺度相关。通过模拟产生不同方差噪声信号进行Contourlet变换，确定各尺度子带系数阈值，得到噪声方差与各尺度子带阈值对应表。对噪声污染图像进行Contourlet变换并估计噪声方差，查表得到各子带阈值，进行系数萎缩处理。实验表明提出的处理方法简单有效，去噪结果具有良好去噪视觉效果和较高峰值信噪比。

Image denoising using nonsubsampled contourlet transform

To remove noise from image a method based on Contourlet Transform was proposed. The construction of Contourlet Transform is based on a nonsubsampled pyramid structure and nonsubsampled Directional Filter Banks. The threshold for coefficients shrinkage was tracked across scales and a library for scale-dependent threshold was built by performing Contourlet Transform on simulated Gaussian noise samples. Images that are corrupted with additive Gaussian noise were de-noised by coefficients shrinkage where threshold was looked up directly in library by estimating standard deviation of noise. The experimental results show that the proposed approach performs effectively in terms of both vision and the PSNR values.