JPEG压缩对相机型号来源取证的影响分析

doi:10.11959/j.issn.2096-109x.2016.00100

Abstract

Abstract:

Focusing on JPEG images transferred by social platform and network media from actual forensic cases,quantities of simulation experiments were carried out to evaluate the performance of three typical methods of model-based source camera identification.The methods are based on the feature of interpolation coefficients of color filter array(CFA),local binary pattern(LBP),and image quality measurement(IQM)with high-order wavelet statis-tic features.The experiments demonstrate that the accuracy decreases rapidly with the compression quality of im-ages decreasing.As a result,methods of model-based source camera identification currently used are demanded bet-ter performance in practical scenes.

Key words: model-based source camera identification, JPEG compression, color filter array, local binary pattern, image quality measures, wavelet transform

CLC Number:

TN911.73

Bo WANG,Jian-feng YIN,Ya-bin LI. Performance evaluation for JPEG compression on model-based source camera identification[J]. Chinese Journal of Network and Information Security, 2016, 2(9): 65-71.

Figures/Tables 11

References 15

[1]	PIVA A . An overview on image forensics[J]. ISRN Signal Process-ing, 2013,(1):1-22.
[2]	王波 . 利用成像引入特征的数字图像被动盲取证研究[D]. 大连: 大连理工大学, 2010.
	WANG B . Study on passive and blind digital image forensics using characteristics introduced by image formation[D]. Dalian: Dalian University of Technology, 2010.
[3]	KOT A C , CAO H . Image and video source class identification[M]// Digital Image Forensics. New York: Springer Press, 2013:157-178.
[4]	SWAMINATHAN A , WU M , LIU K J R . Nonintrusive component forensics of visual sensors using output images[J]. IEEE Transactions on Information Forensics and Security, 2007,2(1):91-105.
[5]	XU G , SHI Y Q . Camera model identification using local binary patterns[C]// IEEE International Confe rence on Multimedia and Expo(ICME). 2012:392-397.
[6]	KHARRAZI M , SENCAR H T , MEMON N . Blind source camera identification[C]// IEEE International Conference on Image Proc-essing(ICIP). 2004:709-712.
[7]	SAN C K , LAM E Y , WONG K K Y . Automatic source cam-eraidentification using the intrinsic lens radial distortion[J]. Optics Express, 2006,14(24):11551-11565.
[8]	王波，孔祥维，尤新钢，等. 基于协方差矩阵的CFA插值盲检测方法[J]. 电子与信息学报, 2009,31(5):1175-1179.
	WANG B , KONG X W , YOU X G , et al. Blind CFA interpolation detection based on covariance matrix[J]. Journal of Electronics and Information Technology, 2009,31(5):1175-1179
[9]	TAN Y , WANG B , LI M , et al. Camera source identification with limited labeled training set[C]// The 14th International Workshop on Digital-forensics and Watermarking,Tokyo. 2015.
[10]	孙雪辉 . 利用多特征和软件成像的手机图像盲取证技术[D]. 大连: 大连理工大学, 2010.
	SUN X H . Blind cell-phone image forensics based on multi- fea-tures and soft-imaging[D]. Dalian: Dalian University of Technology, 2010.
[11]	ISMAIL A . Steganalysis using image quality metrics[J]. IEEE Transactions on Image Processing, 2003,12(2):221-229.
[12]	ESKICOIGLU AM , FISHER P S . Image quality measures and their performance[J]. IEEE Transactions on Communication, 1995,43(12):2959-2965.
[13]	LYU S , FARID H . How realistic is photorealistic[J]. IEEE Transac-tions on Signal Processing, 2005,53(2):845-850.
[14]	WANG B , GUO Y Q , KONG X W , et al. Source camera identifica-tion forensics based on wavelet features[C]// International Confer-ence on Intelligent Information Hiding and Multimedia Signal Processing,Kyoto. 2009:702-705.
[15]	GLOE T , BOHME R . The dresden image database' for benchmark-ing digital image forensics[J]. Journal of Digital Forensic Practice, 2010,3(2-4):150-159.

Metrics

Recommended 0

No Suggested Reading articles found!

特征类别	图像质量特征
	均方误差：Q1；
基于像素差	平均绝对误差：Q2；
	修正无穷范数：Q3
	图像保真度：Q4；
	规范化的交叉相关：Q5；
基于相关性	Czekonowski相关：Q6；平均角度相似性：Q7；
	频谱幅值误差：Q8；
	频谱相位误差：Q9
	频谱相位—幅值误差：Q10；
基于频谱	图像块频谱幅值误差：Q11；图像块频谱相位误差：Q12；
	图像块频谱相位—幅值误差：Q13

序号	相机型号	分辨率	图像数量
Cam1	Casio_EX_Z150	3 264×2 448	350
Cam2	Kodak_M1063	3 664×2 748	350
Cam3	NikonCoolPixS71	4 352×3 264	350
Cam4	Olimpus_mju	3 648×2 736	350
Cam5	Panasonic_DMC	3 648×2 736	350

测试组	最小图像大小/kB	最大图像大小/kB	平均图像大小/kB
原始图像	5 952	5 952	4 227
Q=100	3 012	10 525	6 092
Q=95	1 357	6 588	3 185
Q=90	932	4 839	2 113
Q=85	750	4 250	1 635
Q=80	578	3 492	1 364
Q=75	468	2 799	1 174
Q=70	412	2 643	1 056
Q=65	363	2 451	950

测试组	最小图像大小/kB	最大图像大小/kB	平均图像大小/kB
微信压缩	72	422	208
QQ压缩	38	322	150
微博压缩	192	631	393

		准确率
测试组
测试组
	CFA特征	LBP特征	IQM+高阶小波特征
原始图像	99.4%	98.0%	84.8%
Q=100	98.4%	97.1%	81.4%
Q=95	80.0%	83.4%	74.6%
Q=90	49.0%	52.7%	44.6%
Q=85	32.4%	36.3%	32.4%
Q=80	26.8%	26.5%	25.8%
Q=75	21.2%	23.0%	24.2%
Q=70	21.2%	21.8%	24.4%
Q=65	21.0%	21.0%	23.0%
微信压缩	54.0%	57.2%	53.0%
QQ压缩	23.0%	26.3%	36.0%
微博压缩	95.0%	84.3%	58.0%

Performance evaluation for JPEG compression on model-based source camera identification

RichHTML

PDF下载

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 11

References 15

Related Articles 2

Metrics

Recommended 0

[1]	Wanze CHEN, Liqing HUANG, Jiazhen CHEN, Feng YE, Tianqiang HUANG, Haifeng LUO. Gender forgery of faces by fusing wavelet shortcut connection generative adversarial network [J]. Chinese Journal of Network and Information Security, 2023, 9(3): 150-160.
[2]	Yuguang YANG, Guodong CAO. New image encryption algorithm based on double chaotic system and compressive sensing [J]. Chinese Journal of Network and Information Security, 2022, 8(5): 88-97.