几何约束下噪声一致性的图像拼接篡改检测

doi:10.16180/j.cnki.issn1007-7820.2022.10.009

摘要/Abstract

摘要：

将以图像块为单位的噪声水平估计方法应用在图像篡改定位时,会导致分割边缘呈锯齿状并降低边缘定位准确率。针对该问题,文中提出了一种基于几何约束和噪声一致性分析的图像拼接取证算法。采用基于统计的噪声水平估计与K-means算法对每个图像块实现初步检测定位,提取初步拼接区域边缘的点集合,并以其每个点为中心,依次在边缘图上进行方形范围搜索,拼接区域边缘。随后,利用几何约束筛选算法选择疑似篡改边缘点来定位篡改区域。相较于现有算法,在Columbia上正确检测率相同的情况下,采用文中方法可将错误检测率降低12.7%,并降低算法复杂度。

关键词: 拼接检测, 几何约束, 噪声估计, 图像取证, 边缘检测, 区域定位, K-means, 主成分分析法

Abstract:

When the noise level estimation method based on image block is applied to splicing image localization, it will cause the segmentation edge to be jagged and reduce the accuracy of edge positioning. In view of the problem, this study proposes an image splicing detection algorithm based on geometric constraints and noise consistency analysis. Statistical-based noise level estimation and K-means algorithm are used to achieve preliminary detection and positioning for each image block. The point set at the edge of the initial splicing area is extracted, and each point is used as the center to search for the square range on the edge map in turn to splice the edge of the area. Subsequently, the geometric constraint filtering algorithm is used to select the suspected tampering edge points to locate the tampering area. Compared with the existing algorithm, when the correct detection rate is the same on Columbia, the proposed method can reduce the error detection rate by 12.7% and reduce the complexity of the algorithm.

Key words: splicing detection, geometric constraints, noise estimation, image forensics, edge detection, regional location, K-means, principal component analysis

中图分类号:

TP391.4

路东生,张玉金,朱海,姜月武. 几何约束下噪声一致性的图像拼接篡改检测[J]. 电子科技, 2022, 35(10): 51-58.

LU Dongsheng,ZHANG Yujin,ZHU Hai,JIANG Yuewu. Image-Splicing Forgery Detection Based on Noise Consistency Under Geometric Constraints[J]. Electronic Science and Technology, 2022, 35(10): 51-58.

图/表 10

图1

图2

图3

图4

表1

图5

表2

图6

表3

表4

参考文献 24

[1]	武健, 张大兴, 刘志发. 基于WLD特征的图像复制粘贴篡改检测算法[J]. 电子科技, 2017, 30(7):47-50.
	Wu Jian, Zhang Daxing, Liu Zhifa. An image copy and paste feature tamper detection algorithm based on WLD features[J]. Electronic Science and Technology, 2017, 30(7):47-50.
[2]	王晨璐, 朱婷鸽. 基于伪造痕迹的数字图像盲检测综述[J]. 电子科技, 2015, 28(4):186-188.
	Wang Chenlu, Zhu Tingge. A survey of digital image blind detection based on tampering trace[J]. Electronic Science and Technology, 2015, 28(4):186-188.
[3]	Huang H Y, Ciou A J. Copy-move forgery detection for image forensics using the superpixel segmentation and the Helmert transformation[J]. EURASIP Journal on Image and Video Processing, 2019(1):1-16.
[4]	Chen C C, Lu W Y, Chou C H. Rotational copy-move forgery detection using SIFT and region growing strategies[J]. Multimedia Tools and Applications, 2019, 78(13):18293-18308. doi: 10.1007/s11042-019-7165-8
[5]	熊士婷, 张玉金, 刘婷婷, 等. 基于最优颜色通道的图像拼接检测[J]. 电子科技, 2020, 33(12):49-53.
	Xiong Shiting, Zhang Yujin, Liu Tingting, et al. Image splicing detection based on optimal color channel[J]. Electronic Science and Technology, 2020, 33(12):49-53.
[6]	Moghaddasi Z, Jalab H A, Noor R M. Image splicing forgery detection based on low-dimensional singular value decompo-sition of discrete cosine transform coefficients[J]. Neural Computing and Applications, 2019, 31(11):7867-7877. doi: 10.1007/s00521-018-3586-y
[7]	Kanwal N, Girdhar A, Kaur L, et al. Detection of digital image forgery using fast fourier transform and local fea-tures[C]. Phuket: Proceedings of the International Conference on Automation, Computational and Technology Management, 2019.
[8]	Vega E A A, Fernández E G, Orozco A L S, et al. Passive image forgery detection based on the demosaicing algorithm and jpeg compression[J]. IEEE Access, 2020(8):11815-11823.
[9]	Wu X, Shao Z, Ou P, et al. Application of quantisation-based deep-learning model compression in JPEG image steganaly-sis[J]. The Journal of Engineering, 2018(16):1402-1406.
[10]	Li H, Luo W, Qiu X, et al. Image forgery localization via integrating tampering possibility maps[J]. IEEE Transactions on Information Forensics and Security, 2017, 12(5):1240-1252. doi: 10.1109/TIFS.2017.2656823
[11]	Shen X, Shi Z, Chen H. Splicing image forgery detection using textural features based on the grey level co-occurrence matrices[J]. IET Image Processing, 2017, 11(1):44-53. doi: 10.1049/iet-ipr.2016.0238
[12]	Chen M, Fridrich J, Goljan M, et al. Determining image origin and integrity using sensor noise[J]. IEEE Transactions on Information Forensics and Security, 2008, 3(1):74-90. doi: 10.1109/TIFS.2007.916285
[13]	Lukáŝ J, Fridrich J, Goljan M. Detecting digital image forgeries using sensor pattern noise[C]. San Jose: Proceedings of the Security, Steganography and Watermarking of Multimedia Contents VIII, 2006.
[14]	Popescu A C, Farid H. Statistical tools for digital forensics[C]. Berlin: Proceedings of the International Workshop on Information Hiding, 2004.
[15]	Lyu S, Pan X, Zhang X. Exposing region splicing forgeries with blind local noise estimation[J]. International Journal of Computer Vision, 2014, 110(2):202-221. doi: 10.1007/s11263-013-0688-y
[16]	Mahdian B, Saic S. Using noise inconsistencies for blind image forensics[J]. Image and Vision Computing, 2009, 27(10):1497-1503. doi: 10.1016/j.imavis.2009.02.001
[17]	Pan X, Zhang X, Lyu S. Exposing image splicing with in-consistent local noise variances[C]. Seattle: Proceedings of IEEE International Conference on Computational Photography, 2012.
[18]	Pyatykh S, Hesser J, Zheng L. Image noise level estimation by principal component analysis[J]. IEEE Transactions on Image Processing, 2012, 22(2):687-699. doi: 10.1109/TIP.2012.2221728
[19]	Zeng H, Zhan Y, Kang X, et al. Image splicing localization using PCA-based noise level estimation[J]. Multimedia Tools and Applications, 2017, 76(4):4783-4799. doi: 10.1007/s11042-016-3712-8
[20]	熊士婷, 张玉金, 吴飞, 等. 基于统计噪声水平分析的图像拼接检测[J]. 光电子·激光, 2020, 31(2):214-221.
	Xiong Shiting, Zhang Yujin, Wu Fei, et al. Image splicing detection based on statistical noise level analysis[J]. Journal of Optoelectronics·Laser, 2020, 31(2):214-221.
[21]	Yao H, Wang S, Zhang X, et al. Detecting image splicing based on noise level inconsistency[J]. Multimedia Tools and Applications, 2017, 76(10):12457-12479. doi: 10.1007/s11042-016-3660-3
[22]	Chen G, Zhu F, Ann Heng P. An efficient statistical method for image noise level estimation[C]. Santiago: Proceedings of the IEEE International Conference on Computer Vision, 2015.
[23]	Obaidat M. Creating a complete model of the wooden pattern from laser scanner point clouds using alpha shapes[J]. Jordan Journal of Civil Engineering, 2019, 13(2):269-279.
[24]	Hsu Y F, Chang S F. Detecting image splicing using geome-try invariants and camera characteristics consistency[C]. Shenzhen: Proceedings of the IEEE International Conference on Multimedia and Expositions, 2006.

方法	图像3		图像4		图像5
方法	TPR	FPR	TPR	FPR	TPR	FPR
初步检测结果	0.959 7	0.015 8	0.905 3	0.003 3	0.940 6	0.041 0
最终检测结果	0.996 3	0.009 4	0.939 0	0.003 0	0.995 6	0.040 0

方法	图像6		图像7		图像8
方法	TPR	FPR	TPR	FPR	TPR	FPR
文献[16]	0.864 0	0.022 8	0.725 0	0.024 0	0.612 3	0.061 0
文献[16]-GCF	0.955 5	0.000 0	0.894 0	0.006 2	0.827 0	0.013 3
文献[19]	0.845 7	0.001 5	0.676 2	0.006 6	0.640 2	0.078 1
文献[19]-GCF	0.943 0	0.000 7	0.885 1	0.004 8	0.897 6	0.000 3
本文	0.966 8	0.167 3	0.897 4	0.010 0	0.863 5	0.009 8

方法	无后处理		JPEG QF=95		0.8倍下采样
方法	TPR	FPR	TPR	FPR	TPR	FPR
本文	63.60	13.80	30.40	17.50	64.30	14.10
文献[16]	30.80	21.30	29.60	14.90	36.20	18.90
文献[16]-GCF	32.40	20.40	31.70	14.35	37.70	18.80
文献[17]	36.80	23.00	37.70	20.70	32.00	20.90
文献[19]	47.90	18.50	22.00	10.80	46.00	23.50
文献[19]-GCF	55.70	17.90	21.00	8.60	52.40	24.20
文献[20]	58.50	12.20	29.30	16.80	57.90	13.80
文献[21]	63.60	26.50	52.40	25.00	50.80	25.60

方法	检测时间/s
本文	4.99
文献[16]	0.42
文献[16]-GCF	2.37
文献[17]	4.20
文献[19]	9.42
文献[19]-GCF	11.37
文献[20]	3.04
文献[21]	171.85