基于小波域正则化和贝叶斯规则的图像恢复算法

Image Denosing Based on Wavelet Regularity and Bayesian Regulati on

提出了将小波变换的正则化图像恢复与贝叶斯统计模型分析相结合的方法用于对图像进行消噪处理。正则化图像恢复是条件约束的最优化问题,而小波系数的贝叶斯统计选择是基于图像的随机场观点。两者的有机结合可以辨证地处理正则化参数和算子的选择以及先验模型的分布计算问题。

This work puts forward a new algorithm that the wavelet transformation regularity for image restoration connects with Bayesian statistic modeling, as to make an image denosing. A regularized image restoration is the optimization for some conditional constraint, and the selection of wavelet coefficients based Bayesian statistic is on the image random field view. We could dialectically give the processing about choices of regularity coefficients, operators and the calculation of distributed prior-models according to the two former conditions.