Tampered image recognition based on improved three-stream Faster R-CNN
XU Dai1, YUE Zhang1, YANG Wenxia1, REN Xiao2
1.School of Science, Wuhan University of Technology, WuhanHubei 430070, China
2.School of Automation, Wuhan University of Technology, WuhanHubei 430070, China
A tampered image recognition system with better universality based on convolutional neural network of three-stream feature extraction was proposed to improve the recognition accuracy of three main tampering methods stitching, scaling and rotating, copying and pasting. Firstly, by comparing the similarity of feature sub-blocks according to image local color invariant feature, comparing the noise correlation coefficients of tampered region edges with noise correlation, and calculating the standard deviation contrast of sub-blocks based on image resampling trace, the features of the RGB stream, noise stream and signal stream of the image were extracted separately. Then, through multilinear pooling, combined with an improved piecewise AdaGrad gradient algorithm, the feature dimension reduction and parameter self-adaptive updating were realized. Finally, through network training and classification, three main image tampering methods of stitching, scaling and rotating, copying and pasting were identified and the corresponding tampered areas were located. In order to measure the performance of this model, experiments were carried out on VOC2007 and CIFAR-10 datasets. The experimental results of about 9 000 images show that the proposed model can accurately identify and locate the three tampering methods stitching, scaling and rotating, copying and pasting, and its recognition rates are 0.962,0.956 and 0.935 respectively. Compared with the two-stream feature extraction method in the latest literature, the model has the recognition rates increased by 1.050%, 2.137% and 2.860% respectively. The proposed three-stream model enriches the image feature extraction by convolutional neural network, improves the training performance and recognition accuracy of the network. Meanwhile, controlling the descent speed of parameter learning rate piecewisely by the improved gradient algorithm reduces the over-fitting and convergence oscillation, as well as increases the convergence speed, so as to realize the optimization design of the algorithm.
1 WANG W , DONG J , TAN T . A survey of passive image tampering detection[C]// Proceedings of the 2009 International Workshop on Digital Watermarking, LNCS 5703. Berlin: Springer, 2009:308-322.
2 ZHANG H , SHU H , HAN G N , et al . Blurred image recognition by Legendre moment invariants[J]. IEEE Transactions on Image Processing, 2010, 19(3):596-611.
3 刘一,刘本永 .基于再采样的图像重采样伪作检测[J]. 计算机应用, 2014, 34(3):815-819. (LIU Y, LIU B Y. Image resampling artifact detection based on resampling[J]. Journal of Computer Applications, 2014, 34(3):815- 819.)
4 王春华,韩栋 . 基于区域直方图和特征相关匹配规则的图像复制-粘贴篡改检测算法[J]. 电子测量与仪器学报, 2018, 32(4):103-109. (WANG C H, HAN D. Algorithm for image copy-paste forgery detection based on region histogram and feature correlation matching rule[J]. Journal of Electronic Measurement and Instrumentation, 2018, 32(4): 103-109.)
5 及莹,陆月明 . 一种基于SIFT算法的图像镜像变换识别方法[J]. 计算机应用研究, 2013, 30(3):924-925, 941. (JI Y, LU Y M. Method based on SIFT for recognition of flip images[J]. Application Research of Computers, 2013, 30(3):924-925, 941.)
6 郑继明,崔玉岩,耿金玲,等 . 运用DWT和ORB的图像篡改检测方法[J]. 计算机工程与应用, 2017, 53(11): 187-191. ZHENG J M , CUI Y Y , GENG J L , et al . Image tampering detection method using DWT and ORB[J]. Computer Engineering and Applications, 2017, 53(11): 187-191.
7 BAYAR B , STAMM M C . A deep learning approach to universal image manipulation detection using a new convolutional layer[C]// Proceedings of the 4th ACM Workshop on Information Hiding and Multimedia Security. New York: ACM, 2016:5-10.
8 常亮,邓小明,周明全,等 . 图像理解中的卷积神经网络[J]. 自动化学报, 2016, 42(9):1300-1312. CHANG L , DENG X M , ZHOU M Q , et al . Convolution neural networks in image understanding[J]. Acta Automatica Sinica, 2016, 42(9): 1300-1312.
9 刘万军,梁雪剑,曲海成 . 基于双重优化的卷积神经网络图像识别算法[J]. 模式识别与人工智能, 2016, 29(9):856-864. (LIU W J, LIANG X J, QU H C. Convolutional neural network algorithm based on double optimization for image recognition[J]. Pattern Recognition and Artificial Intelligence, 2016, 29(9):856-864.)
10 张文达,许悦雷,倪嘉成,等 . 基于多尺度分块卷积神经网络的图像目标识别算法[J]. 计算机应用, 2016, 36(4):1033-1038 ZHANG W D , XU Y L , NI J C , et al . Image target recognition based on multi-scale block convolution neural network[J]. Journal of Computer Applications, 2016, 36(4):1033-1038.
11 曲长文,刘晨,周强,等 . 基于分块 CNN 的多尺度SAR图像目标分类算法[J]. 雷达科学与技术, 2018, 16(2):169-173,180. QU C W , LIU C , ZHOU Q , et al . Multi-scale SAR image target classification algorithm based on block convolutional neural network[J]. Radar Science and Technology, 2018, 16(2):169-173,180.
12 REN S , HE K , GIRSHICK R , et al . Faster R-CNN: towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6):1137-1149.
13 ZHOU P , HAN X , MORARIU V I , et al . Learning rich features for image manipulation detection[C]// Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2018:1053-1061.
14 亚春,武金岗,王净,等 . 小波变换图像融合规则性能的分析研究[J].计算机工程与应用, 2010, 46(8):180-182, 189. YA C, WU J G , WANG J , et al . Analysis on image fusion rules based on wavelet transform[J]. Computer Engineering and Applications, 2010, 46(8):180-182, 189.
15 刘淑春 . 基于RGB颜色特征和小波变换的图像复制-粘贴篡改检测[D]. 昆明:云南大学, 2011. (LIU S C. The copy-move forgery detection based on RGB color feature and wavelet transform[D]. Kunming: Yunnan University, 2011.)
16 洪拥筠,刘本永 . 基于图像背景噪声相关性的篡改检测[J]. 贵州大学学报(自然科学版), 2015, 32(3):93-96, 118. (HONG Y J, LIU B Y. Image forgery detection using correlation of background noise[J]. Journal of Guizhou University (Natural Science), 2015, 32(3): 93-96, 118.)
17 姚恒,魏为民,唐振军 . 重采样图像的盲检测技术[J]. 计算机工程与应用, 2010, 46(30):166-168, 187. (YAO H, WEI W M, TANG Z J. Survey of digital forensics technology for image resampling detection[J]. Computer Engineering and Applications, 2010, 46(30): 166-168, 187.)
18 左菊仙,邓坚 . 基于重采样痕迹的图像伪造检测[J]. 计算机应用与软件, 2016, 33(10):328-333. (ZUO J X, DENG J. Image forgery detection based on resampling traces[J]. Computer Applications and Software, 2016, 33(10): 328-333.)
19 骆伟祺,黄继武,丘国平 . 鲁棒的区域复制图像篡改检测技术[J]. 计算机学报, 2007, 30(11):1998-2007. (LUO W Q, HUANG J W, QIU G P. Robust detection of region-duplication forgery in digital image[J]. Chinese Journal of Computers, 2007, 30(11): 1998-2007.)
20 刘帆,刘鹏远,张峻宁,等 . 一种改进的深度学习模型自适应学习率策略[J]. 兵工自动化, 2019, 38(1):72-77. LIU F , LIU P Y , ZHANG J N , et al . An improved depth learning model adaptive learning rate strategy[J]. Ordnance Industry Automation, 2019, 38(1):72-77.