改进的SIFT耦合特征点集群的图像伪造检测算法
|
摘要
针对当前图像伪造检测算法进行图像伪造检测时主要通过设定比例阀值来实现特征匹配,存在检测误差大、鲁棒性不强等不足,提出了改进的SIFT耦合特征点集群的图像伪造检测算法.首先,采用二进小波变换提取伪造图像的低频子带以用于特征点检测;然后,基于特征点邻域旋转不变纹理特性,改进了SIFT机制,生成新的特征描述子对其进行描述,减少误匹配,并提出了自适应匹配策略,通过搜索最优比例阀值,以提高算法检测精度及鲁棒性;最后,通过构建特征点的均值漂移向量,对特征点均值和特征点的偏差进行度量,实现特征点的集群,从而完成图像的伪造检测.仿真结果显示:跟当前的伪造检测方法相比,本文方法具有更高的检测精度与鲁棒性,呈现出较好的ROC特性.
In view of the current image forgery detection algorithm for image forgery detection,the ratio of the threshold has mainly been set to achieve feature matching,which lead to the detection error is big;the robustness is not strong and so on.A effective image forgery detection algorithm based on improved SIFT coupled feature point clustering has been proposed in this paper.Firstly,the low frequency sub bands of the forged image are extracted by using dyadic wavelet transform to detect the feature points.Then,using feature point neighborhood rotation invariant texture characteristics to describe the feature points,formation characteristic descriptor to reduce the false matching,adaptive matching strategy is proposed,search the optimal ratio of the threshold in order to improve the accuracy and robustness of detecting algorithm,to improve the traditional SIFT feature descriptor generation and matching strategies.Finally,the mean shift vector of feature points is constructed to measure the deviation of feature points and feature points to realize the clustering of feature points and the image forgery detection is completed.The simulation results show that compared with the current forgery detection methods;this method has the characteristics of small detection error,strong robustness and so on.
In view of the current image forgery detection algorithm for image forgery detection,the ratio of the threshold has mainly been set to achieve feature matching,which lead to the detection error is big;the robustness is not strong and so on.A effective image forgery detection algorithm based on improved SIFT coupled feature point clustering has been proposed in this paper.Firstly,the low frequency sub bands of the forged image are extracted by using dyadic wavelet transform to detect the feature points.Then,using feature point neighborhood rotation invariant texture characteristics to describe the feature points,formation characteristic descriptor to reduce the false matching,adaptive matching strategy is proposed,search the optimal ratio of the threshold in order to improve the accuracy and robustness of detecting algorithm,to improve the traditional SIFT feature descriptor generation and matching strategies.Finally,the mean shift vector of feature points is constructed to measure the deviation of feature points and feature points to realize the clustering of feature points and the image forgery detection is completed.The simulation results show that compared with the current forgery detection methods;this method has the characteristics of small detection error,strong robustness and so on.
引文
[1]欧红玉,陈曦,宋燕辉.基于LBP的图像复制篡改检测[J].计算机应用与软件,2013,9(30):170-172,178.
[2]高铁杠,杨富圣,盛国瑞.一种新的基于DCT域系数对直方图的图像篡改取证方法[J].光电子·激光,2014,11(25):2196-2202.
[3]KALRA G S,TALWAR R,SADAWARTI H.Adaptive Digital Image Watermarking for Color Images in Frequency Domain[J].Multimedia Tools and Applications,2014,3(36):416-423.
[4]KAUSHIKA R,BAJAJB R K,MATHEWC J I.On Image Forgery Detection Using Two Dimensional Discrete Cosine Transform and Statistical Moments[J].Procedia Computer Science,2015,1(70):130-136.
[5]REZA D,KHASHAYAR Y,SAEED M.Copy-Move Forgery Detection Using Multiresolution Local Binary Patterns[J].Forensic Science International,2013,1(231):61-72.
[6]王森,伍星,刘韬.基于二进小波变换的多尺度图切割方法[J].计算机工程与应用,2015,13(51):9-14.
[7]张瑞芳,程晓辉,宋子航.融合灰色马尔科夫理论的二进小波图像的复制-粘贴篡改检测算法[J].桂林理工大学学报,2014,4(34):775-781.
[8]MOHAMMAD F H,VIJAY A,AVINAS G.Copy-move Image Forgery Detection Using an Efficient and Robust Method Combining Un-decimated Wavelet Transform and Scale Invariant Feature Transform[J].AASRI Procedia,2014,3(9):84-91.
[9]AMERINI I,BARNI M,CALDELLI R.Removal and Injection of Keypoints for SIFT-Based Copy-Move Counter-Forensics[J].EURASIP Journal on Information Security,2013,5(1):132-141.
[10]MARYAM J,GEORGE B,MUHAMMAD H.Accurate and Robust Localization of Duplicated Region in Copy-Move Image Forgery[J].Machine Vision And Applications,2014,2(25):451-475.
[11]SWAPNIL H K,AVINASH D G.Copy-Move Attack Forgery Detection by Using SIFT[J].International Journal of Innovative Technology and Exploring Engineering,2013,2(5):221-224.
[12]杜振龙,焦丽鑫,李晓丽.帧内复制粘贴视频伪造的盲检测[J].中国图象图形学报,2014,6(9):825-834.
[13]PETER M.Semi-Supervised Kernel Mean Shift Clustering.[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2014,6(36):1201-1215.
[14]HE Hai-zhen,HUANG Xin-hang,KUANG Jun.Exposing Copy-Move Forgeries Based on a Dimension-reduced Sift Method[J].Information Technology Journal,2013,14(12):2975-2979.
[15]GAVIN L,FRANK Y S,LIAO Hong-yuan.An Efficient Expanding Block Algorithm for Image Copy-Move Forgery Detection[J].Information Sciences,2012,4(23):253-265.
[16]张智丰,裴志利.基于模糊局部二值模式算子的图像伪造检测[J].计算机工程与设计,2014,33(12):3284-3290.
[17]ISAAC M M,WILSCY M.Image Forgery Detection Based on Gabor Wavelets and Local Phase Quantization[J].Procedia Computer,2012,58(12):76-83.
[2]高铁杠,杨富圣,盛国瑞.一种新的基于DCT域系数对直方图的图像篡改取证方法[J].光电子·激光,2014,11(25):2196-2202.
[3]KALRA G S,TALWAR R,SADAWARTI H.Adaptive Digital Image Watermarking for Color Images in Frequency Domain[J].Multimedia Tools and Applications,2014,3(36):416-423.
[4]KAUSHIKA R,BAJAJB R K,MATHEWC J I.On Image Forgery Detection Using Two Dimensional Discrete Cosine Transform and Statistical Moments[J].Procedia Computer Science,2015,1(70):130-136.
[5]REZA D,KHASHAYAR Y,SAEED M.Copy-Move Forgery Detection Using Multiresolution Local Binary Patterns[J].Forensic Science International,2013,1(231):61-72.
[6]王森,伍星,刘韬.基于二进小波变换的多尺度图切割方法[J].计算机工程与应用,2015,13(51):9-14.
[7]张瑞芳,程晓辉,宋子航.融合灰色马尔科夫理论的二进小波图像的复制-粘贴篡改检测算法[J].桂林理工大学学报,2014,4(34):775-781.
[8]MOHAMMAD F H,VIJAY A,AVINAS G.Copy-move Image Forgery Detection Using an Efficient and Robust Method Combining Un-decimated Wavelet Transform and Scale Invariant Feature Transform[J].AASRI Procedia,2014,3(9):84-91.
[9]AMERINI I,BARNI M,CALDELLI R.Removal and Injection of Keypoints for SIFT-Based Copy-Move Counter-Forensics[J].EURASIP Journal on Information Security,2013,5(1):132-141.
[10]MARYAM J,GEORGE B,MUHAMMAD H.Accurate and Robust Localization of Duplicated Region in Copy-Move Image Forgery[J].Machine Vision And Applications,2014,2(25):451-475.
[11]SWAPNIL H K,AVINASH D G.Copy-Move Attack Forgery Detection by Using SIFT[J].International Journal of Innovative Technology and Exploring Engineering,2013,2(5):221-224.
[12]杜振龙,焦丽鑫,李晓丽.帧内复制粘贴视频伪造的盲检测[J].中国图象图形学报,2014,6(9):825-834.
[13]PETER M.Semi-Supervised Kernel Mean Shift Clustering.[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2014,6(36):1201-1215.
[14]HE Hai-zhen,HUANG Xin-hang,KUANG Jun.Exposing Copy-Move Forgeries Based on a Dimension-reduced Sift Method[J].Information Technology Journal,2013,14(12):2975-2979.
[15]GAVIN L,FRANK Y S,LIAO Hong-yuan.An Efficient Expanding Block Algorithm for Image Copy-Move Forgery Detection[J].Information Sciences,2012,4(23):253-265.
[16]张智丰,裴志利.基于模糊局部二值模式算子的图像伪造检测[J].计算机工程与设计,2014,33(12):3284-3290.
[17]ISAAC M M,WILSCY M.Image Forgery Detection Based on Gabor Wavelets and Local Phase Quantization[J].Procedia Computer,2012,58(12):76-83.