摘要
基于小波分解的图像小波系数在层内和层间解相关而相互依存的客观现实,提出了一个联合层内和层间两方向系数的非高斯联合概率分布模型.以此模型作为先验分布,在Bayesian估计理论的框架下,导出小波系数闭式的最大后验(MAP)估计公式,并用高斯噪声污染的典型图像进行了实验.结果显示,由该估计公式计算得到的去噪图像不仅有较少的均方误差(MSE),还具有保护和增强边缘的能力.
Based on the inter-and intra-scale coefficients' decorrelating but also the dependent properties of wavelet-based decomposed image, a new local non-Gaussian joint probability distribution model is proposed, and following that, a new closed maximum a posteriori(MAP) estimating formula is derived under the Bayesian estimation theory by using this model as the prior distribution model. At last, several numerical examples are given, the experiments show the denoised images have not only a lower mean-square error(MSE) ,but also a better ability of edge preservation and enhancement.
出处
《北京理工大学学报》
EI
CAS
CSCD
北大核心
2005年第4期356-359,共4页
Transactions of Beijing Institute of Technology
基金
国家自然科学基金资助项目(60472110)
关键词
图像去噪
小波系数模型
概率分布
image denoising
wavelet coefficients model
probability distribution
