图像小波系数的高斯混合模型研究

图像小波系数的统计分布具有非高斯特性，可以用高斯混合模型进行描述。提出了一种随像素自适应调整的混合高斯模型，每个系数建模为两个均值为零、方差不同的正态分布之和，利用局部贝叶斯阈值对小波系数进行分类，通过当前系数邻域窗中两类系数的信息，得到大、小方差以及有关概率的模型参数估计。将此模型应用于图像去噪，根据贝叶斯后验均值估计理论设计了Wiener滤波器。通过与三种代表性去噪算法的比较实验，表明了这种基于模型的滤波算法的有效性。

Study of Gaussian mixture model for image wavelet coefficients

A pixel-adaptive Gaussian mixture model was proposed in which each coefficient was a mixture of two normal distributions with the same zero mean value and different variance. Wavelet coefficients were classified into two categories using local Bayesian threshold, and the model parameters such as large and small variances, related probabilities, could be estimated from the information of the two classified coefficients in a neighbourlng window. The model was applied to image denoising, and Wiener filter was designed according to Bayesian posterior mean estimation theory. Comparative simulation results with three representative denoising algorithms demonstrate that this model-based filtering algorithm is effective.