摘要： 利用图像小波子带内系数的相关性，提出了一种局部自适应小波去噪方法。首先在贝叶斯最大后验概率准则下推导出基于拉普拉斯先验分布的MAP估计表达式和子带MapShrink阈值。为得到局部自适应的MapShrink阈值和去噪算法，提出将子带内的每个小波系数建模为具有不同边缘标准差的拉普拉斯分布，而边缘标准差又假设为强局部相关的随机变量，可通过邻域局部窗口进行估计。实验结果表明，与经典的子带自适应去噪算法相比较，该方法获得了明显的峰值信噪比增益，主观视觉效果也得到了改善。

关键词: 小波系数, 图像去噪, 拉普拉斯模型, 最大后验估计, MapShrink阈值

Abstract: A locally adaptive wavelet denoising method is proposed by exploiting the correlation among image wavelet coefficients in a sub-band. Firstly, under the rule of Bayesian maximum a posteriori (MAP), this paper deduces Laplacian prior distribution based MAP estimator formula and sub-band adaptive MapShrink threshold. In order to make this threshold locally adaptive, a new stochastic model for wavelet coefficients is presented, in which each coefficient in a sub-band is assumed to be Laplacian with different marginal standard deviations, and these marginal standard deviations are modeled as random variables with high local correlation and thus can be estimated from a local neighborhood. Experiment results demonstrate that compared with classical sub-band adaptive algorithms, the proposed denoising method has significantly increased peak signal-to-noise ratio and improved the quality of subjective visual effect.

Key words: Wavelet coefficients, Image denoising, Laplacian model, Maximum aposteriori estimation, MapShrink threshold

中图分类号: 

• zil@scuec.edu.cn