改进的基于小波双变量模型的图像去噪算法

An Improved Image Denoising Algorithm Based on Wavelet Bivariate Model

研究图像小波系数间的统计相关性并建立适当的模型,可以显著提高图像处理的质量。在贝叶斯最大后验估计理论框架下,讨论了Sendur提出的双变量模型,用MAP估计方法推导了对应的萎缩函数,分析了基于双变量模型去噪算法的不足,在此基础上进行了改进,利用MAP软阈值对第L级三个高频子带进行局部自适应处理。实验结果表明了改进后算法的有效性。

The quality of image processing can be significantly improved by exploiting the correlation among image wavelet coefficients and modeling the statistics for these coefficients.Under the framework of Bayesian MAP estimation theory,the bivafiate model presented by Sendur is investigated,and the corresponding shrinkage function is derived by MAP estimator.The paper pointes out some drawbacks of this bivariate model based on denoising algorithm and proposes an improved method,in which the coefficients of three high frequency sub-bands in the L-th wavelet decomposition level are modified by MAP soft thresholding rule via locally adaptive fashion.The validity of the proposed method is demonstrated by experiment results.