摘要
针对数字图像来源判别准确率不高的问题,提出了基于模式噪声的图像来源认证方法。该算法首先利用小波降噪算法从已拍摄图像中提取模式噪声,然后求取模式噪声的小波和马尔可夫特征,最后采用支持向量机(support vector machine,SVM)分类方法,利用噪声特征组合的分类器进行图像来源分类,并对分类结果进行定性和定量分析。实验结果表明,该方法对不同品牌和同一品牌不同型号的相机都有较高的判别准确率。
In order to improve the accuracy of source identification of digital image,this paper put forward the image source authentication method based on pattern noise.Firstly,the algorithm extracted the noise which had taken from the image,using the wavelet de-noising algorithm.Then it calculated the features of wavelet and Markov,and finally used the support vector machine,by the combination of the noise characteristics as a classifier to indentify source image.At the same time,it analysed the classified results qualitatively and quantitatively.The experimental results demonstrate that in allusion to different brand and different models of the same brand,the method has a high accuracy.
出处
《计算机应用研究》
CSCD
北大核心
2014年第6期1918-1920,共3页
Application Research of Computers
基金
重庆市基础与前沿研究计划资助项目(cstc2013jcyjA40034)
重庆市教委科学技术研究资助项目(KJ130528)
重庆市科委重点实验室专项经费资助项目
