基于高斯径向基函数的复制-移动篡改检测算法
|
摘要
为提高对模糊、噪声、压缩、旋转等篡改区域的检测精度与鲁棒性,提出基于离散余弦变换与Gaussian径向基函数的复制-移动篡改检测算法。将色彩图像变换为灰度图像,将其划分为一定大小的重叠子块;对每个字块进行离散余弦变换(discrete cosine transform,DCT),利用DCT分量表示子块;根据DCT分量构建图像的系数矩阵,引入一个高斯径向基函数(Gaussian radial basis function,RBF)核PCA(principal components analysis,PCA),将其系数矩阵的特征值作为块的特征表示;计算不同块之间的Euclidean距离作为不同区域的相似性,并与给定的阈值比较,完成伪造检测。实验结果表明,与当前篡改检测技术相比,所提算法具有更好的检测精度和效率。
To improve the detection accuracy and robustness of fuzzy,noise,compression,rotation and other tampered regions,a copy move tamper detection algorithm based on DCT and Gaussian radial basis functions was proposed.The color images were transformed into grayscale images,and they were divided into overlapping blocks of a certain size.Discrete cosine transform was used for each block,the DCT component was represented as a block.The coefficients matrix of the image was constructed according to the DCT component,to extract block feature accurately,the feature dimension was reduced,a Gaussian radial basis function kernel function was introduced,and the eigenvalues of the coefficient matrix were expressed as blocks.The Euclidean distance between different blocks was calculated as the similarity of different regions,the forgery detection was completed by comparing with the given threshold.Experimental results show that the proposed algorithm has good detection accuracy and efficiency compared with the current popular tamper detection.
To improve the detection accuracy and robustness of fuzzy,noise,compression,rotation and other tampered regions,a copy move tamper detection algorithm based on DCT and Gaussian radial basis functions was proposed.The color images were transformed into grayscale images,and they were divided into overlapping blocks of a certain size.Discrete cosine transform was used for each block,the DCT component was represented as a block.The coefficients matrix of the image was constructed according to the DCT component,to extract block feature accurately,the feature dimension was reduced,a Gaussian radial basis function kernel function was introduced,and the eigenvalues of the coefficient matrix were expressed as blocks.The Euclidean distance between different blocks was calculated as the similarity of different regions,the forgery detection was completed by comparing with the given threshold.Experimental results show that the proposed algorithm has good detection accuracy and efficiency compared with the current popular tamper detection.
引文
[1]LI Weizheng,BAI Guangwei,TU Xiumei.Image forgery detection algorithm based on local phase quantization[J].Computer Engineering and Design,2017,38(2):449-452(in Chinese).[李伟征,白光伟,涂秀梅.基于局部相位量化的图像伪造检测算法[J].计算机工程与设计,2017,38(2):449-452.]
[2]Lee JC,Chang CP,Chen WK.Detection of copy-move image forgery using histogram of orientated gradients[J].Information Sciences,2015,321(12):250-262.
[3]Li L,Li S,Zhu H.Detecting copy-move forgery under affine transforms for image forensics[J].Computers and Electrical Engineering,2014,40(6):1951-1962.
[4]JIAO Lixin,DU Zhenlong.Copy-paste image forgery blind detection based on mean shift[J].Journal of Computer Applications,2014,34(3):806-809(in Chinese).[焦丽鑫,杜振龙.基于均值漂移的图像复制粘贴伪造盲检测[J].计算机应用,2014,34(3):806-809.]
[5]XIA Lei,ZHOU Bing.Copy-move image forgery detection algorithm based on pixel prediction error coupled likelihood mapping[J].Chinese Journal of Quantum Electronics,2016,33(2):153-161(in Chinese).[夏蕾,周冰.像素预测误差耦合似然映射的移动-复制图像伪造检测算法[J].量子电子学报,2016,33(2):153-161.]
[6]Mahdian B,Saic S.Detection of copy-move forgery using a method based on blur moment invariants[J].Forensic Science international,2015,17121(3):180-189.
[7]CHEN Rui,YANG Haigang, WANG Fei.Design of multistandard discrete cosine transform based on coarse-grained reconfigurable array[J].Journal of Electronics&Information Technology,2015,37(1):206-213(in Chinese).[陈锐,杨海钢,王飞.基于粗粒度可重构阵列结构的多标准离散余弦变换设计[J].电子与信息学报,2015,37(1):206-213.]
[8]SHANG Jingjing,SUN Liujie, WANG Wenju.Holographic digital blind watermark algorithm for 3D point cloud model based on discrete cosine transform[J].Packaging Engineering,2015,36(13):111-114(in Chinese).[商静静,孙刘杰,王文举.基于离散余弦变换的三维点云模型全息盲水印[J].包装工程,2015,36(13):111-114.]
[9]LU Qiuju,GUO Tianyin,REN Minhong.Face recognition in illumination condition based on discrete cosine transform and principal components analysis[J].Laser Journal,2015,36(4):126-130(in Chinese).[鲁秋菊,郭天印,任民宏.基于离散余弦变换和主成分分析的光照条件人脸识别[J].激光杂志,2015,36(4):126-130.]
[10]Mahmood T, Mahmood MT,Shah M.Copy-move forgery detection technique for forensic analysis in digital images[J].Hindawi Publishing Corporation,2016,13(2):1-13.
[11]ZHAO Rentao,GUO Caiqiao,LI Huade.Adaptive low contrast image enhancement algorithm based on the RBF neural network[J].Journal of Graphics,2015,36(3):432-437(in Chinese).[赵仁涛,郭彩乔,李华德.基于RBF神经网络的低对比度图像自适应增强算法[J].图学学报,2015,36(3):432-437.]
[12]YANG Zhixiang,SHI Wenhao,ZHANG Ning.Mufti-objective optimization algorithm based on lexicographic forting and solution association[J].Journal of Chinese Computer Systems,2016,37(9):2024-2028(in Chinese).[杨志翔,史文浩,张宁.基于字典排序和解关联的多目标优化算法[J].小型微型计算机系统,2016,37(9):2024-2028.]
[13]LIANG Qiaoping,XU Yongjian,YANG Baohong,et al.Segmentation of the ink layer in z-direction cross-section views of offset paper by using Euclidean distance and digital microscope[J].Packaging Engineeing,2015,36(17):126-130(in Chinese).[梁巧萍,徐永建,杨保宏,等.基于Euclidean距离的纸张z向截面中墨层图像分割[J].包装工程,2015,36(17):126-130.]
[14]SHI Wen,ZHU Xuefang,ZHU Guang.Adaptive edge detection algorithm of MRI image based on morphology[J].Chinese Journal of Scientific Instrument,2013,34(2):408-414(in Chinese).[师文,朱学芳,朱光.基于形态学的MRI图像自适应边缘检测算法[J].仪器仪表学报,2013,34(2):408-414.]
[15]ZHANU Feiyan,LI Junfeng,SHEN Junmin.Human behavior recognition based on the statistical characteristics of gradient and optical flow[J].Journal of Optoelectronics·Laser,2015,26(8):1593-1601(in Chinese).[张飞燕,李俊峰,沈军民.基于梯度和光流统计特性的人体行为识别[J].光电子·激光,2015,26(8):1593-1601.]
[16]WANG Dawei,LYU Ailong,CHEN Meijie.A 3Dface data preprocessing method based on CASIA 3D[J].Journal of Shanxi Normal University(Natural Science Edlition),2015,29(4):32-36(in Chinese).[王大为,吕爱龙,陈美婕.基于CASIA 3D的三维人脸数据预处理方法[J].山西师范大学学报(自然科学版),2015,29(4):32-36.]
[17]LI Jingfu,ZHANG Fei.Study on image forgery detection algorithm based on convex optimization mechanism coupled sensor pattern noise[J].Computer Measurement&Control,2015,23(5):1678-1681(in Chinese).[李景富,张飞.凸优化耦合传感器模式噪声的图像伪造检测[J].计算机测量与控制,2015,23(5):1678-1681.]
[2]Lee JC,Chang CP,Chen WK.Detection of copy-move image forgery using histogram of orientated gradients[J].Information Sciences,2015,321(12):250-262.
[3]Li L,Li S,Zhu H.Detecting copy-move forgery under affine transforms for image forensics[J].Computers and Electrical Engineering,2014,40(6):1951-1962.
[4]JIAO Lixin,DU Zhenlong.Copy-paste image forgery blind detection based on mean shift[J].Journal of Computer Applications,2014,34(3):806-809(in Chinese).[焦丽鑫,杜振龙.基于均值漂移的图像复制粘贴伪造盲检测[J].计算机应用,2014,34(3):806-809.]
[5]XIA Lei,ZHOU Bing.Copy-move image forgery detection algorithm based on pixel prediction error coupled likelihood mapping[J].Chinese Journal of Quantum Electronics,2016,33(2):153-161(in Chinese).[夏蕾,周冰.像素预测误差耦合似然映射的移动-复制图像伪造检测算法[J].量子电子学报,2016,33(2):153-161.]
[6]Mahdian B,Saic S.Detection of copy-move forgery using a method based on blur moment invariants[J].Forensic Science international,2015,17121(3):180-189.
[7]CHEN Rui,YANG Haigang, WANG Fei.Design of multistandard discrete cosine transform based on coarse-grained reconfigurable array[J].Journal of Electronics&Information Technology,2015,37(1):206-213(in Chinese).[陈锐,杨海钢,王飞.基于粗粒度可重构阵列结构的多标准离散余弦变换设计[J].电子与信息学报,2015,37(1):206-213.]
[8]SHANG Jingjing,SUN Liujie, WANG Wenju.Holographic digital blind watermark algorithm for 3D point cloud model based on discrete cosine transform[J].Packaging Engineering,2015,36(13):111-114(in Chinese).[商静静,孙刘杰,王文举.基于离散余弦变换的三维点云模型全息盲水印[J].包装工程,2015,36(13):111-114.]
[9]LU Qiuju,GUO Tianyin,REN Minhong.Face recognition in illumination condition based on discrete cosine transform and principal components analysis[J].Laser Journal,2015,36(4):126-130(in Chinese).[鲁秋菊,郭天印,任民宏.基于离散余弦变换和主成分分析的光照条件人脸识别[J].激光杂志,2015,36(4):126-130.]
[10]Mahmood T, Mahmood MT,Shah M.Copy-move forgery detection technique for forensic analysis in digital images[J].Hindawi Publishing Corporation,2016,13(2):1-13.
[11]ZHAO Rentao,GUO Caiqiao,LI Huade.Adaptive low contrast image enhancement algorithm based on the RBF neural network[J].Journal of Graphics,2015,36(3):432-437(in Chinese).[赵仁涛,郭彩乔,李华德.基于RBF神经网络的低对比度图像自适应增强算法[J].图学学报,2015,36(3):432-437.]
[12]YANG Zhixiang,SHI Wenhao,ZHANG Ning.Mufti-objective optimization algorithm based on lexicographic forting and solution association[J].Journal of Chinese Computer Systems,2016,37(9):2024-2028(in Chinese).[杨志翔,史文浩,张宁.基于字典排序和解关联的多目标优化算法[J].小型微型计算机系统,2016,37(9):2024-2028.]
[13]LIANG Qiaoping,XU Yongjian,YANG Baohong,et al.Segmentation of the ink layer in z-direction cross-section views of offset paper by using Euclidean distance and digital microscope[J].Packaging Engineeing,2015,36(17):126-130(in Chinese).[梁巧萍,徐永建,杨保宏,等.基于Euclidean距离的纸张z向截面中墨层图像分割[J].包装工程,2015,36(17):126-130.]
[14]SHI Wen,ZHU Xuefang,ZHU Guang.Adaptive edge detection algorithm of MRI image based on morphology[J].Chinese Journal of Scientific Instrument,2013,34(2):408-414(in Chinese).[师文,朱学芳,朱光.基于形态学的MRI图像自适应边缘检测算法[J].仪器仪表学报,2013,34(2):408-414.]
[15]ZHANU Feiyan,LI Junfeng,SHEN Junmin.Human behavior recognition based on the statistical characteristics of gradient and optical flow[J].Journal of Optoelectronics·Laser,2015,26(8):1593-1601(in Chinese).[张飞燕,李俊峰,沈军民.基于梯度和光流统计特性的人体行为识别[J].光电子·激光,2015,26(8):1593-1601.]
[16]WANG Dawei,LYU Ailong,CHEN Meijie.A 3Dface data preprocessing method based on CASIA 3D[J].Journal of Shanxi Normal University(Natural Science Edlition),2015,29(4):32-36(in Chinese).[王大为,吕爱龙,陈美婕.基于CASIA 3D的三维人脸数据预处理方法[J].山西师范大学学报(自然科学版),2015,29(4):32-36.]
[17]LI Jingfu,ZHANG Fei.Study on image forgery detection algorithm based on convex optimization mechanism coupled sensor pattern noise[J].Computer Measurement&Control,2015,23(5):1678-1681(in Chinese).[李景富,张飞.凸优化耦合传感器模式噪声的图像伪造检测[J].计算机测量与控制,2015,23(5):1678-1681.]