数字图像盲取证技术的一些研究
|
摘要
随着计算机技术的发展,使用photoshop这种功能强大的软件来修改数字图像不再是什么困难的事情。然而任何事情都具有两面性,目前图像造假事件不断的涌现使得人们对数字媒体的信赖完全抹杀,数字媒体的公信力严重降低。那些可以作为司法鉴定,案件证据,关键信息取证的数码照片已经丢失了它往日应有的效力。这导致必然会带来一些牵涉到诸如法律取证中的图像真实性、图像媒体的版权、个人的隐私保护等问题。数字图像取证技术已经成为了现今社会的一个重大课题。
本文首先论述了数字图像盲取证技术的研究的基本框架,现有方法的分类。然后从JPEG双压缩篡改检测和自然图像与计算机图像鉴别两个方向进行了深入的研究,一方面借鉴前人的研究成果,一方面提出了自己的新方法,并通过实验证明了所提方法的有效性。论文的主要工作包括以下几个方面:
(1)JPEG压缩参考图像的引入:本文提出并实现了一种基于参考图像的双JPEG压缩的数字图像篡改的检测算法。引入参考图像的目的是使参考图像JPEG图片与未经过篡改前的JPEG具有相同的压缩特性。经过双重压缩的篡改图像经过将待检测的图像删除左侧4列,再次进行压缩时,打乱了原有的JPEG图像的DCT块使得再次进行压缩后的JPEG图片不具备双重压缩的统计特性。
(2)提出ΔE曲线概念:本文提出了ΔE这一概念来具体实现JPEG双压缩篡改检测,并取得了不错的效果。ΔE是一系列压缩图像与原图像的失真程度的变化率。这一概念的提出对本文定位JPEG篡改位置起到关键性作用。这一变化率可以更容易反映篡改位置图像的双压缩情况。原始图像经过JPEG双压缩的ΔE均值曲线波动比较巨烈,在背景图像首次压缩的质量因子的位置ΔE有明显的符号转变。而参考图像单次JPEG压缩的ΔE均值曲线较为平缓。
(3)基于预测误差图像的像素相关一致性的自然图像与计算机图像的鉴别:本文提出了增加了预测误差图像可以把这种相关性信息提取出来。预测误差图像的优点是提取的各个特征之间相关性更小,泛化能力更强。分析和总结了这些数字图像自然数据特性的研究意义和方法。确定适合的颜色空间:本文通过大量实验证明,HSV颜色空间在提取像素特征进行分类时比其它颜色空间可以获得更好的效果。
(4)提出一类和多类分类器结合鉴别图像类别的思想:本文还提出了一种一类和多类分类器结合的方法对图像的类别进行分类。并对多类分类器区分自然图像,计算机三维图像,计算机二维卡通图像进行了实验,并取得了良好的效果。
With the development of computer technology, the use of photoshop can easily modify the pictures. However, everything has two sides. Many people began to fake photos. This makes people no longer trust in digital media. And Digital media's credibility has been seriously reduced. The effectiveness of digital photos has been lost. This leads inevitably lead to some problems, such as Image authenticity, image media, copyright, personal privacy protection. Digital image forensics technology has become a major issue in contemporary society.
This paper discusses the basic framework for study of digital forensics based on image content. From the double-compressed JPEG tamper detection and identification of natural images and computer graphics in both directions in-depth study. The main thesis work includes the following aspects:
(1) Reference to the calibration of compressed JPEG images: Propose a calibration image based on double JPEG compressed digital image tampering detection algorithm. After a double compression detection of tampering with the image after the image are to be removed on the left 4, When the compression again, this disrupted the original JPEG image of the DCT block. And again compressed JPEG images do not have the statistical properties of double-compressed.
(2) The introduction of the concept ofΔE curve:ΔE is a distortion level of the rate of change in a series of compressed images and between the original images. The rate of change can more easily reflect the altered location of the image pairs of compression. The original image after JPEG compressionΔE pairs of relatively strong fluctuations in the mean curve. This is a clear sign of change inΔE of Background image.
(3) Prediction error image based on the related to consistency of pixels of natural images and computer graphics of the identification: This increased the prediction error image. It can be extracted from the correlation of information. Prediction error image has greater generalization ability.Determine the color space: Through experiments HSV color space in the pixel feature extraction have better results.
(4) Propose OC-SVM and MC-SVM combination with identifying the image type of thinking: Through experiments, the use of MC-SVM achieved good results.
本文首先论述了数字图像盲取证技术的研究的基本框架,现有方法的分类。然后从JPEG双压缩篡改检测和自然图像与计算机图像鉴别两个方向进行了深入的研究,一方面借鉴前人的研究成果,一方面提出了自己的新方法,并通过实验证明了所提方法的有效性。论文的主要工作包括以下几个方面:
(1)JPEG压缩参考图像的引入:本文提出并实现了一种基于参考图像的双JPEG压缩的数字图像篡改的检测算法。引入参考图像的目的是使参考图像JPEG图片与未经过篡改前的JPEG具有相同的压缩特性。经过双重压缩的篡改图像经过将待检测的图像删除左侧4列,再次进行压缩时,打乱了原有的JPEG图像的DCT块使得再次进行压缩后的JPEG图片不具备双重压缩的统计特性。
(2)提出ΔE曲线概念:本文提出了ΔE这一概念来具体实现JPEG双压缩篡改检测,并取得了不错的效果。ΔE是一系列压缩图像与原图像的失真程度的变化率。这一概念的提出对本文定位JPEG篡改位置起到关键性作用。这一变化率可以更容易反映篡改位置图像的双压缩情况。原始图像经过JPEG双压缩的ΔE均值曲线波动比较巨烈,在背景图像首次压缩的质量因子的位置ΔE有明显的符号转变。而参考图像单次JPEG压缩的ΔE均值曲线较为平缓。
(3)基于预测误差图像的像素相关一致性的自然图像与计算机图像的鉴别:本文提出了增加了预测误差图像可以把这种相关性信息提取出来。预测误差图像的优点是提取的各个特征之间相关性更小,泛化能力更强。分析和总结了这些数字图像自然数据特性的研究意义和方法。确定适合的颜色空间:本文通过大量实验证明,HSV颜色空间在提取像素特征进行分类时比其它颜色空间可以获得更好的效果。
(4)提出一类和多类分类器结合鉴别图像类别的思想:本文还提出了一种一类和多类分类器结合的方法对图像的类别进行分类。并对多类分类器区分自然图像,计算机三维图像,计算机二维卡通图像进行了实验,并取得了良好的效果。
With the development of computer technology, the use of photoshop can easily modify the pictures. However, everything has two sides. Many people began to fake photos. This makes people no longer trust in digital media. And Digital media's credibility has been seriously reduced. The effectiveness of digital photos has been lost. This leads inevitably lead to some problems, such as Image authenticity, image media, copyright, personal privacy protection. Digital image forensics technology has become a major issue in contemporary society.
This paper discusses the basic framework for study of digital forensics based on image content. From the double-compressed JPEG tamper detection and identification of natural images and computer graphics in both directions in-depth study. The main thesis work includes the following aspects:
(1) Reference to the calibration of compressed JPEG images: Propose a calibration image based on double JPEG compressed digital image tampering detection algorithm. After a double compression detection of tampering with the image after the image are to be removed on the left 4, When the compression again, this disrupted the original JPEG image of the DCT block. And again compressed JPEG images do not have the statistical properties of double-compressed.
(2) The introduction of the concept ofΔE curve:ΔE is a distortion level of the rate of change in a series of compressed images and between the original images. The rate of change can more easily reflect the altered location of the image pairs of compression. The original image after JPEG compressionΔE pairs of relatively strong fluctuations in the mean curve. This is a clear sign of change inΔE of Background image.
(3) Prediction error image based on the related to consistency of pixels of natural images and computer graphics of the identification: This increased the prediction error image. It can be extracted from the correlation of information. Prediction error image has greater generalization ability.Determine the color space: Through experiments HSV color space in the pixel feature extraction have better results.
(4) Propose OC-SVM and MC-SVM combination with identifying the image type of thinking: Through experiments, the use of MC-SVM achieved good results.
引文
[1]. Stefan K.,Fabien A.P.P.吴秋新等译.信息隐藏技术一隐写术与数字水印[M].北京:人民邮电出版社,2001: 15-17.
[2]. Ingemar J.Cox,Matthew L.Miller,and Jeffrey A.Bloom著,王颖,黄志蓓等译.数字水印[M].北京::电子工业出版社,2003.23-24.
[3].周琳娜,数字图像盲取证技术研究[D],硕士,北京:北京邮电大学.2007:11-12.
[4]. NgT T,Chang S F,Lin C Y, Qibin Sun Q B.Passive-blind image forensies[C]. Multimedia Security Technologies for Digital Rights. New York: Elsevier,2006:17-28.
[5]. Lanh T V,Chong K S,Emmanuel S,Kankanhalli M S. A survey on digital camera image forensic methods[C] . Proceedings of 2007 IEEE International Confereneeon Multimedia and Expo.Beijing,China: IEEE,2007:16-19.
[6]. Senear H T,Memon N. Overview of state-of-the-art in digital image forensics[C]. Part of Indian Statistical Institute Platinum Jubilee Monograph Series Titled S tatistical Science and lnterdisciplinary Researh. USA: World Scientific Press,2008:175-178.
[7].吴琼,李国辉,涂丹,孙韶杰.面向真实性鉴别的数字图像盲取证技术综述[J]自动化学报2008,34(12):1458-1466.
[8]. Willam B Pennebaker,Joan L Mitchell著.黎洪松,成实译.JPEG静止图像数据压缩标准[S].北京:学苑出版社,1996.
[9]. Fridrich J,Soukal D,Lukas J.Deteetion of copy-move forgery in digital images [C]. Proeeedings of Digital Forensic Researeh WbrkshoP. Cleveland,USA: Springer,2003.1-10.
[10]. Popeseu A C ,Farid H. Exposing Digital Forgeries by Deteeting Duplicated Image Regions[C],Technical Report TR2004-515,Department of Computer Science,Dartmouth College,2004.
[11]. Li G H,Wu Q,Tu D.Sun S J. A .sorted neighborhood approach for detecting duplicated regions in image forgeries based on DWT and SVD[C]. Proceedings of 2007 IEEE International Conference on Multimedia and Expo.Beijing:IEEE,2007:1750-1753.
[12]. Fridrich J,Lukas J Estimation of Primary quantization matrix in double Compressed JPEG images[C]. In: Proeeedings of Digital Forensie Researeh workshoP. Cleveland USA: IEEE,2003.
[13]. Popeseu A C.Statistical Tools for Digital Forensies[Ph.D.Dissertation],Department of Computer Science,Dartmouth College,2005.
[14]. Dai Meng,Lin Jia-Jun,Mao Jia-Fa.The analysis and detection of double JPEG compression [J]. Journal of Image and Grrohics,2006,11(11):1619-1622.
[15]. H J F,Lin Z C,Wang L F,Tang X O. Detecting doctored JPEG images via DCT coeffieient analysis[C].Proceedings of 9th EuroPean Conferenee on Computer Vision,Grajz,Austria: Springer,2006:423-435.
[16]. Popeseu A C,Farid H.Exposing digital forgeries bydetceting traees of resampling[J],IEEE Transactions on Signal Processing,2005,53(2):758-767.
[17]. ZhuXiu-Ming,XuanGuo-Rong, Yao Qiu-Ming,Tong Xue-Feng,ShiYun-Qing. Resampling detection in information forensies[J]. Journal o f Computer Applications. 2006,26(11):2596-2597.
[18]. Lyu S W ,Farid H. How realistie is Photoreajistie?[C] IEEE transactions on signal Processing 2005,53(2):845-850.
[19]. Lyu S W. Natural Image Statistics for Digital Image Forensies[ph. D. dissertation],Department of ComputerScience,Darthmouth College,2005.
[20]. Aveibas I, Memon N,Sankur B. Steganalysis using image quality metries[C]. IEEE transactions on Image Processing,2003,12(2): 221-229.
[21]. Aveibasl , MemonN , SankurB , Ramkumar M. A classifier design for detecting image manipulation[C].Proceeding of 2004 International Conference on Image Processing. SingaPore ,Singapore:IEEE,2004:2645-2648.
[22]. Bayram S,Avcibas I,Sankur B,Memon N. Image manipulation detection with binary similarity measures[C]. Proceedings of the 13th EuroPean Signal Proeessing Conferenee.Antalya,Turkey: Curran Assoeiates,2005:752-755.
[23]. Bayram S,Avcibas I,Sankur B,Memon N.Image manipulation detection[J].Journal of Electronie imaging 2006,15(4):1117-1138.
[24].崔霞,童学锋,宣国荣,黄聪.基于直方图频域矩的自然图像和计算机图形的鉴别[C].第七届全国信息隐藏暨多媒体安全学术大会会议论文集.2007:276-279.
[25].张益真.刘滔编著.Visual C++实现MPEG/JPEG编解码技术[M].北京:北京邮电出版社,2002::1-110.
[26]. G.K.Wallace.The JPEG still picture compression standard[S].IEEE Transactions on Consumer Electronics, 1991,34(4):30-44.
[27].罗俊松.基于JPEG2000标准的数字图像压缩的算法研究[D].成都理工大学,硕士,2003:29-33.
[28]. Fan Z.and de Queiroz R.L.Identification of Bitmap Compression History:JPEG Detection and Quantizer Estimation[J]. IEEE Transactions on Image Processing, 2003 12(2),89-90.
[29].方敏.JPEG算法的研究与实现[D].辽宁工程技术大学,硕士,2002:1-5.
[30]. A.C.Popescu and H.Farid.Statistical Tools for Digital Forensics[R].6th International Workshop on Information Hiding,Toronto,Canada,2004.
[31]. R.C.Reininger and J.D.Gibson.Distributions of the two-dimensional DCT coefficients for images[C].IEEE Trans.Commun., 1983:835-839.
[32]. E.Y.Lam and J.W.Goodman.A mathematical analysis of the DCT coefficient distributions for images[C].IEEE Trans.Image2000.10:1661-1666.
[33]. S.R.Smoot and L.A.Rowe.Laplacian model for AC DCT terms in image and video coding[C].Proc.9th IEEEImage and Multidimension Digital Signal Processing Workshop,1996.
[34].于雪燕.一种基于双JPEG压缩的数字图像的篡改检测方法[D].上海师范大学,硕士,2007:10-12
[35]. POPESCU A C, FARID H. Statistical tools for digital forensics[C] Proceedings of the 6th Information Hiding Workshop, 2004.
[36]. FARID H. DetectingDigitalForgeriesUsingBispectralAnalysis[R]. TechnicalReportAIM-1657, Massachusetts Institue ofTechnology, 1999.
[37]. Popescu Alin C,Statistical Tools for Digital Forensics[Ph.D.dissertation],Department of Computer Science,Darmouth College,2005.
[38].戴蒙,林家骏,毛家发.JPEG二次压缩的分析与检测[J],中国图象图形学报,2006,11(11):1619~1622.
[39]. A.C.Popescu and H.Farid,Statistical Tools for Digital Forensics[C],6th InternationalWorkshop on Informational Hiding,128-147,Toronto,Canada,2004.
[40].李晟,张新鹏,利用JPEG压缩特性的合成图像检测[J],应用科学学报,2008.,26(3):281-287,
[41]. Pevny T, Fridrich J, Determining the Stego Algorithm for JPEG Images[C], Special Issue of IEE Proceedings - Information Security, 2006, 153(3):75-139.
[42].郑二功,平西建.一种基于相邻像素一致性的数码照片与计算机图形鉴别方法[J].计算机研究与发展. 2009,46(Suppl), :258-262.
[43]. Kohavi.R,John,G.H.Wrappers for Feature Subset Selection[C]. Artificial Intelligence, 1997:273 - 3240.
[44]. Mladernnic.D,Grobelnik.M.Feature Selection for unbalanced class distribution and Na?ve Bayes[R].In Proceedings of the Sixteenth International Conference on Machine teaming. Bled,Morgan Kaufmann,1999:258-267.
[45].李晶皎,朱志良,王爱侠等译.Sergios Theodoridis等著.模式识别(第二版)[M].电子工业出版社,2004:106-202.
[46].边肇祺,张学工等编著.模式识别(第二版)[M].北京:清华大学出版社,2000:296-303.
[47].关新平编著.信号处理与模式式别[M].北京:机械工业出版社,2006:58-64.
[48].冈萨雷斯.数字图像处理(第二版) [M].电子工业出版社,2007:70-80.
[49].杨关良,李忠杰,徐小杰.基于代表色的图像检索方法研究[J].仪器仪表学报.2003,24(4):411-413.
[50].卢燕斌.监控视频中运动目标检测和追踪的研究及应用[D],上海交通大学.硕士2007.
[51].王波,孔祥维,付海燕.联合OC-SVM和MC-SVM的图像来源取证方法[J].计算机研究与发展.2009.9:1456-1461.
[2]. Ingemar J.Cox,Matthew L.Miller,and Jeffrey A.Bloom著,王颖,黄志蓓等译.数字水印[M].北京::电子工业出版社,2003.23-24.
[3].周琳娜,数字图像盲取证技术研究[D],硕士,北京:北京邮电大学.2007:11-12.
[4]. NgT T,Chang S F,Lin C Y, Qibin Sun Q B.Passive-blind image forensies[C]. Multimedia Security Technologies for Digital Rights. New York: Elsevier,2006:17-28.
[5]. Lanh T V,Chong K S,Emmanuel S,Kankanhalli M S. A survey on digital camera image forensic methods[C] . Proceedings of 2007 IEEE International Confereneeon Multimedia and Expo.Beijing,China: IEEE,2007:16-19.
[6]. Senear H T,Memon N. Overview of state-of-the-art in digital image forensics[C]. Part of Indian Statistical Institute Platinum Jubilee Monograph Series Titled S tatistical Science and lnterdisciplinary Researh. USA: World Scientific Press,2008:175-178.
[7].吴琼,李国辉,涂丹,孙韶杰.面向真实性鉴别的数字图像盲取证技术综述[J]自动化学报2008,34(12):1458-1466.
[8]. Willam B Pennebaker,Joan L Mitchell著.黎洪松,成实译.JPEG静止图像数据压缩标准[S].北京:学苑出版社,1996.
[9]. Fridrich J,Soukal D,Lukas J.Deteetion of copy-move forgery in digital images [C]. Proeeedings of Digital Forensic Researeh WbrkshoP. Cleveland,USA: Springer,2003.1-10.
[10]. Popeseu A C ,Farid H. Exposing Digital Forgeries by Deteeting Duplicated Image Regions[C],Technical Report TR2004-515,Department of Computer Science,Dartmouth College,2004.
[11]. Li G H,Wu Q,Tu D.Sun S J. A .sorted neighborhood approach for detecting duplicated regions in image forgeries based on DWT and SVD[C]. Proceedings of 2007 IEEE International Conference on Multimedia and Expo.Beijing:IEEE,2007:1750-1753.
[12]. Fridrich J,Lukas J Estimation of Primary quantization matrix in double Compressed JPEG images[C]. In: Proeeedings of Digital Forensie Researeh workshoP. Cleveland USA: IEEE,2003.
[13]. Popeseu A C.Statistical Tools for Digital Forensies[Ph.D.Dissertation],Department of Computer Science,Dartmouth College,2005.
[14]. Dai Meng,Lin Jia-Jun,Mao Jia-Fa.The analysis and detection of double JPEG compression [J]. Journal of Image and Grrohics,2006,11(11):1619-1622.
[15]. H J F,Lin Z C,Wang L F,Tang X O. Detecting doctored JPEG images via DCT coeffieient analysis[C].Proceedings of 9th EuroPean Conferenee on Computer Vision,Grajz,Austria: Springer,2006:423-435.
[16]. Popeseu A C,Farid H.Exposing digital forgeries bydetceting traees of resampling[J],IEEE Transactions on Signal Processing,2005,53(2):758-767.
[17]. ZhuXiu-Ming,XuanGuo-Rong, Yao Qiu-Ming,Tong Xue-Feng,ShiYun-Qing. Resampling detection in information forensies[J]. Journal o f Computer Applications. 2006,26(11):2596-2597.
[18]. Lyu S W ,Farid H. How realistie is Photoreajistie?[C] IEEE transactions on signal Processing 2005,53(2):845-850.
[19]. Lyu S W. Natural Image Statistics for Digital Image Forensies[ph. D. dissertation],Department of ComputerScience,Darthmouth College,2005.
[20]. Aveibas I, Memon N,Sankur B. Steganalysis using image quality metries[C]. IEEE transactions on Image Processing,2003,12(2): 221-229.
[21]. Aveibasl , MemonN , SankurB , Ramkumar M. A classifier design for detecting image manipulation[C].Proceeding of 2004 International Conference on Image Processing. SingaPore ,Singapore:IEEE,2004:2645-2648.
[22]. Bayram S,Avcibas I,Sankur B,Memon N. Image manipulation detection with binary similarity measures[C]. Proceedings of the 13th EuroPean Signal Proeessing Conferenee.Antalya,Turkey: Curran Assoeiates,2005:752-755.
[23]. Bayram S,Avcibas I,Sankur B,Memon N.Image manipulation detection[J].Journal of Electronie imaging 2006,15(4):1117-1138.
[24].崔霞,童学锋,宣国荣,黄聪.基于直方图频域矩的自然图像和计算机图形的鉴别[C].第七届全国信息隐藏暨多媒体安全学术大会会议论文集.2007:276-279.
[25].张益真.刘滔编著.Visual C++实现MPEG/JPEG编解码技术[M].北京:北京邮电出版社,2002::1-110.
[26]. G.K.Wallace.The JPEG still picture compression standard[S].IEEE Transactions on Consumer Electronics, 1991,34(4):30-44.
[27].罗俊松.基于JPEG2000标准的数字图像压缩的算法研究[D].成都理工大学,硕士,2003:29-33.
[28]. Fan Z.and de Queiroz R.L.Identification of Bitmap Compression History:JPEG Detection and Quantizer Estimation[J]. IEEE Transactions on Image Processing, 2003 12(2),89-90.
[29].方敏.JPEG算法的研究与实现[D].辽宁工程技术大学,硕士,2002:1-5.
[30]. A.C.Popescu and H.Farid.Statistical Tools for Digital Forensics[R].6th International Workshop on Information Hiding,Toronto,Canada,2004.
[31]. R.C.Reininger and J.D.Gibson.Distributions of the two-dimensional DCT coefficients for images[C].IEEE Trans.Commun., 1983:835-839.
[32]. E.Y.Lam and J.W.Goodman.A mathematical analysis of the DCT coefficient distributions for images[C].IEEE Trans.Image2000.10:1661-1666.
[33]. S.R.Smoot and L.A.Rowe.Laplacian model for AC DCT terms in image and video coding[C].Proc.9th IEEEImage and Multidimension Digital Signal Processing Workshop,1996.
[34].于雪燕.一种基于双JPEG压缩的数字图像的篡改检测方法[D].上海师范大学,硕士,2007:10-12
[35]. POPESCU A C, FARID H. Statistical tools for digital forensics[C] Proceedings of the 6th Information Hiding Workshop, 2004.
[36]. FARID H. DetectingDigitalForgeriesUsingBispectralAnalysis[R]. TechnicalReportAIM-1657, Massachusetts Institue ofTechnology, 1999.
[37]. Popescu Alin C,Statistical Tools for Digital Forensics[Ph.D.dissertation],Department of Computer Science,Darmouth College,2005.
[38].戴蒙,林家骏,毛家发.JPEG二次压缩的分析与检测[J],中国图象图形学报,2006,11(11):1619~1622.
[39]. A.C.Popescu and H.Farid,Statistical Tools for Digital Forensics[C],6th InternationalWorkshop on Informational Hiding,128-147,Toronto,Canada,2004.
[40].李晟,张新鹏,利用JPEG压缩特性的合成图像检测[J],应用科学学报,2008.,26(3):281-287,
[41]. Pevny T, Fridrich J, Determining the Stego Algorithm for JPEG Images[C], Special Issue of IEE Proceedings - Information Security, 2006, 153(3):75-139.
[42].郑二功,平西建.一种基于相邻像素一致性的数码照片与计算机图形鉴别方法[J].计算机研究与发展. 2009,46(Suppl), :258-262.
[43]. Kohavi.R,John,G.H.Wrappers for Feature Subset Selection[C]. Artificial Intelligence, 1997:273 - 3240.
[44]. Mladernnic.D,Grobelnik.M.Feature Selection for unbalanced class distribution and Na?ve Bayes[R].In Proceedings of the Sixteenth International Conference on Machine teaming. Bled,Morgan Kaufmann,1999:258-267.
[45].李晶皎,朱志良,王爱侠等译.Sergios Theodoridis等著.模式识别(第二版)[M].电子工业出版社,2004:106-202.
[46].边肇祺,张学工等编著.模式识别(第二版)[M].北京:清华大学出版社,2000:296-303.
[47].关新平编著.信号处理与模式式别[M].北京:机械工业出版社,2006:58-64.
[48].冈萨雷斯.数字图像处理(第二版) [M].电子工业出版社,2007:70-80.
[49].杨关良,李忠杰,徐小杰.基于代表色的图像检索方法研究[J].仪器仪表学报.2003,24(4):411-413.
[50].卢燕斌.监控视频中运动目标检测和追踪的研究及应用[D],上海交通大学.硕士2007.
[51].王波,孔祥维,付海燕.联合OC-SVM和MC-SVM的图像来源取证方法[J].计算机研究与发展.2009.9:1456-1461.