图像高斯噪声估计的极大似然算法  

Maximum Likelihood Estimation Algorithm for Gaussian Noise Image

作　　者：王蓓[1] 张根耀[1] 李智[1] 

机构地区：[1]延安大学数学与计算机科学学院,陕西延安716000

出　　处：《计算机技术与发展》2014年第7期153-156,共4页Computer Technology and Development

基　　金：陕西省教育科研项目(2010JK904);陕西省教改项目(2013JK1124);2012年延安大学研究生教育计划创新项目

摘　　要：提出一种图像高斯噪声极大似然估计方法,目的是估计出噪声图像所含噪声大小。首先,根据高斯噪声模型的特点,用极大似然法估计噪声值,对图像所含噪声模型进行分析。其次,把噪声图像用直方图表示,从归一化直方图中选出不同的样本观测值,用极大似然算法对噪声的方差进行估计。最后,用MATLAB对该方法进行了模拟实验,实验结果表明此方法所得的图像噪声的方差与实际图像噪声的方差近似相等。所以,此方法无论是在准确性上还是在可行性上均具有优良的特性。A maximum likelihood estimation method for Gaussian noise image is provided in this paper, the purpose is to estimate the noise size noise image contains. Firstly according to the characteristics of Gaussian noise model,using the maximum likelihood method to estimate the noise value,the image contains noise models are analyzed. Secondly,the noise figure that expressed with a histogram and dif-ferent sample observations from the normalized histogram are selected,with the maximum likelihood algorithm to estimate the noise vari-ance. Finally,the MATLAB simulation experiment is carried out,the results demonstrate that the image noise variance the method obtains with the actual image noise variance is given approximately equal. Therefore,this method in both accuracy and feasibility are excellent.

关 键 词：极大似然估计 白高斯噪声 噪声图像 噪声估计 直方图 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]