非均匀噪声图像的小波去噪  

Wavelet Denoising of Images Corrupted by Non-uniform Noise

作　　者：张凌霜[1] 赵玉娟[1] 李秀英[1] 

机构地区：[1]西安电子科技大学理学院数学系,陕西西安710071

出　　处：《现代电子技术》2005年第9期39-41,共3页Modern Electronics Technique

摘　　要：目前的小波阈值去噪一般考虑单幅均匀噪声样本,针对非均匀噪声模型,提出基于邻域估计法的逐点阈值去噪算法;并将多幅均匀噪声样本的联合去噪法推广到非均匀噪声模型下。其中,阈值函数的选取是:基于信号方差和噪声方差的自适应的逐点Bayes阈值;对多幅非均匀噪声样本采用阈值去噪与加权平均结合的方案,加权系数由各象素点的噪声方差确定,并比较了图像域和变换域两种加权方法的性能。Most methods assume that the noise is spatially stationary additive white Gaussian noise (AWGN).We aim at noise that is spatially nonspatially stationary AWGN′s,and give pointwise thresholding method based on the neighborhood estimations in this text;then extend joint wavelet denoising method using multiple copies corrupted by spatially stationary AWGN′s to nonspatially stationary case.Image thresholding is selfregulating Bayes function which base on noise variances and signal variances.We combine thresholding and average method for multiple copies corrupted by nonspatially stationary AWGN′s,which weighted value is reckoned by noise variances,and compare two different pointwise weighted averages in image and wavelet domain.

关 键 词：逐点噪声方差估计 逐点Bayes阈值 逐点加权平均 非均匀噪声 

分 类 号：TN911.73[电子电信—通信与信息系统]