-
摘要:
传统图像拼接检测算法通过研究人员手动构造拼接特征,随着科技的进步以及图像处理技术的不断发展,手动构造特征的局限性逐渐体现出来,鲁棒性较弱,位置不易确定等。为了解决这些问题,构建了一种卷积神经网络(CNN),将卷积核前置并固定,自主学习相关特征从而检测拼接篡改的图像区域。经过一系列研究,发现拼接篡改图像的拼接篡改区域特征可以被CNN模型学习。在CNN模型之前,卷积核使用高通滤波器,激活函数采用指数线性单元(ELU),使得CNN模型具有识别拼接篡改图像边缘痕迹等特征的能力。检测结果表明:在IEEE IFS-TC图像拼接取证竞赛训练集上对拼接篡改图像拼接篡改区域定位的准确率为84.3%,对拼接篡改区域判定的真负类率为96.18%。
-
关键词:
- 卷积神经网络(CNN) /
- 图像拼接取证 /
- 深度学习 /
- 指数线性单元(ELU) /
- 损失函数
Abstract:The traditional image stitching detection algorithm manually constructs the stitching features by researchers. With the advancement of technology and the continuous development of image processing technology, the limitations of the features of manual construction, such as weak robustness and difficult positioning, are gradually manifested. Aimed at this kind of problem, this paper proposes to construct a Convolutional Neural Network (CNN) by means of fixed pre-convolution kernel, and detect the image tampering area by feature self-learning. Through experiments and research, it is found that the features of the mosaic tampering area of the spliced tamper image can be learned by the CNN model. Prior to the CNN model, the convolution kernel uses a high-pass filter and the activation function uses an Exponential Linear Unit (ELU), which makes the CNN model be capable of identifying features such as splicing and tampering with image edge traces. The detection results show that the positioning accuracy for the falsification image's tampering area is 84.3% in the IEEE IFS-TC image forensics training set and the detection true negative rate of the tampering area is 96.18%.
-
表 1 不同分类器在测试集上的实验结果
Table 1. Experimental results of different classifiers on test sets
% 分类器 准确率 召回率 真负类率 B_CNN 82.9 70.32 95.54 R_CNN 82.1 69.82 95.15 G_CNN 81.9 69.85 93.95 RGB_CNN集成分类器 84.3 73.84 96.18 -
[1] HAGHIGHI B B, TAHERINIA A H, MOHAJERZADEH A H.TRLG:Fragile blind quad watermarking for image tamper detection and recovery by providing compact digests with optimized quality using LWT and GA[J].Information Sciences, 2019, 486:204-230. doi: 10.1016/j.ins.2019.02.055 [2] 李燕, 钟磊, 李健.基于LBP和共生矩阵的图像拼接篡改检测[J].武汉大学学报(理学版), 2015, 61(6):517-524. http://d.old.wanfangdata.com.cn/Periodical/whdxxb-zr201506003LI Y, ZHONG L, LI J.Detection of image splicing forgery based on LBP and co-occurrence matrix[J].Journal of Wuhan University(Natural Science Edition), 2015, 61(6):517-524(in Chinese). http://d.old.wanfangdata.com.cn/Periodical/whdxxb-zr201506003 [3] LUO W, QU Z, HUANG J, et al.A novel method for detecting cropped and recompressed image block[C]//IEEE International Conference on Acoustics, Speech and Signal Processing.Piscataway, NJ: IEEE Press, 2007: 217-220. [4] SWAMINATHAN A, WU M, LIU K J R.Component forensics of digital cameras: A non-intrusive approach[C]//2006 40th Annual Conference on Information Sciences and Systems.Piscataway, NJ: IEEE Press, 2007: 1194-1199. [5] ZHANG Q, LU W, WANG R, et al.Digital image splicing detection based on Markov features in block DWT domain[J].Multimedia Tools and Applications, 2018, 77(23):31239-31260. doi: 10.1007/s11042-018-6230-z [6] KANG X, LI Y, QU Z, et al.Enhancing source camera identification performance with a camera reference phase sensor pattern noise[J].IEEE Transactions on Information Forensics and Security, 2012, 7(2):393-402. doi: 10.1109/TIFS.2011.2168214 [7] 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 [8] 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 [9] HAN J G, PARK T H, MOON Y H, et al.Quantization-based Markov feature extraction method for image splicing detection[J].Machine Vision and Applications, 2018, 29(3):543-552. doi: 10.1007/s00138-018-0911-5 [10] BELFERDI W, BEHLOUL A, NOUI L.A Bayer pattern-based fragile watermarking scheme for color image tamper detection and restoration[J].Multidimensional Systems and Signal Processing, 2019, 30(3):1093-1112. doi: 10.1007/s11045-018-0597-x [11] RAJPUT V, ANSARI I A.Image tamper detection and self-recovery using multiple median watermarking[J/OL].Multimedia Tools and Applications, 2019: 1-17[2019-03-01].https://doi.org/10.1007/s11042-019-07971-w. [12] SHIN H C, ORTON M R, COLLINS D J, et al.Stacked autoencoders for unsupervised feature learning and multiple organ detection in a pilot study using 4D patient data[J].IEEE Transactions on Pattern Analysis & Machine Intelligence, 2013, 35(8):1930-1943. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=b739bde4251b7ad954d369b069d501bf [13] VINCENT P, LAROCHELLE H, LAJOIE I, et al.Stacked denoising autoencoders:Learning useful representations in a deep network with a local denoising criterion[J].Journal of Machine Learning Research, 2010, 11(12):3371-3408. http://cn.bing.com/academic/profile?id=1b5321cd3b5fdc0bc4d9b0d4fa8975ef&encoded=0&v=paper_preview&mkt=zh-cn [14] LUKAS J, FRIDRICH J, GOLJAN M.Digital camera identification from sensor pattern noise[J].IEEE Transactions on Information Forensics and Security, 2006, 1(2):205-214. doi: 10.1109/TIFS.2006.873602 [15] LECUN Y, BOTTOU L, BENGIO Y, et al.Gradient-based learning applied to document recognition[J].Proceedings of the IEEE, 1998, 86(11):2278-2324. doi: 10.1109/5.726791