基于Cauchy分布模型与NSST变换的图像去噪算法

Image denoising algorithm based on Cauchy distribution model and NSST transform

非下采样Shearlet变换(NSST)具有良好的方向敏感性,各向异性以及平移不变性,是接近最优的多尺度稀疏表示方法.提出一种基于先验柯西(Cauchy)模型的NSST域图像去噪方法,利用Cauchy分布对NSST变换域子带系数概率分布进行拟合,作为先验分布模型,再通过最大后验概率(MAP)方法估计不含噪声的系数.该方法不但保留了传统统计模型去噪方法中的优点,还通过对NSST具有更好拟合效果的柯西分布模型作为先验的概率分布模型,使估计出的系数更接近于原始图像的系数.大量仿真实验验证了所提出方法的有效性.

In order to solve the problem that the multi-scale transform threshold denoising method does not consider the correlation between sub-band coefficients,a denoising method based on statistical model is proposed.The non-downsampling Shearlet transform（NSST）has a good directional sensitivity,anisotropy and translation invariance,which is close to the optimal multi-scale sparse representation.In this paper,the effectiveness of the Cauchy distribution model as a priori probability model is analyzed.An image denoising algorithm based on the Cauchy distribution model and NSST transform is proposed.The statistical model denoising method based on wavelet and NSST transform based on Laplacian distribution Denoising method are compared.The simulation results show that the presently proposed method has better denoising effect.