摘要
图像拼接是图像篡改中最基础和最常用的方法。针对图像拼接的操作,提出一种基于图像统计特性的图像拼接检测方法。在图像的离散余弦变换域上,提取马尔可夫特征来表征图像的统计信息,并采取监督学习的方法,用支持向量机进行分类,识别出图像是否被篡改。为了寻求最优的分类结果,采用径向基函数为核函数,并用10折交叉验证的方法,得出适合的参数C和G。实验结果表明:在阈值T=3和分块大小N=2,4,8的组合时,准确率达到最优。
Image Mosaic is one of the most basic and common methods.Aiming at the operation of image Mosaic,an image Mosaic detection method based on image statistical characteristics is proposed.In the discrete cosine transform domain of image,Markov features are extracted to represent the statistical information of the image,and a supervised learning method is adopted to classify and identify whether the image is tampered with or not by using support vector machine.In order to find the best classification results,this paper uses the radial basis function as the Kernel function,and uses the 10 fold cross validation method to search for the appropriate parameters C and G.The experimental results show that the optimal accuracy is achieved when threshold T=3 and block size N=2,4 and 8.
作者
宋凯
贾先欣
SONG Kai;JIA Xian-xin(Shenyang Ligong University,Shenyang 110159,Liaoning Province)
出处
《沈阳工程学院学报（自然科学版）》
2022年第4期78-81,共4页
Journal of Shenyang Institute of Engineering:Natural Science
基金
辽宁省自然科学基金指导计划项目(2019-ZD-0260)。
关键词
图像拼接
离散余弦变换域
马尔可夫特征
支持向量机
10折交叉验证
image Mosaic
discrete cosine transform domain
Markov features
support vector machine
10 fold cross validation
