A Deep Learning Based Digital Forensic Solution to Blind Source Identification of Facebook Images
- 642 Downloads
Source Camera Identification is a digital forensic way of attributing a contentious image to its authentic source, especially used in legal application domains involving terrorism, child pornography etc. The state–of–the–art source camera identification techniques, however, are not suitable to work with images downloaded from online social networks. This is because, online social networks impart specific image artefacts, due to proprietary image compression requirements for storage and transmission, which prevents accurate forensic source investigations. Moreover, each social network has its own compression standards, which are never made public due to ethical issues. This makes source identification task even more difficult for forensic analysts. In present day and age, where there is abundant use of social networks for image transmission, it is high time that source camera identification with images downloaded from social networks, be efficiently addressed. In this paper, we propose a deep learning based digital forensic technique for source camera identification, on images downloaded from Facebook. The proposed deep learning technique is adapted from the popular ResNet50 network, which majorly consists of convolutional layers and a few pooling layers. Our experimental results prove that the proposed technique outperforms the traditional source camera identification methods.
KeywordsCamera model identification Classification Deep learning Facebook ResNet Source camera identification
This work is funded by Council of Scientific and Industrial Research (CSIR), Govt. of India, Grant No. 22(0736)/17/EMR-II, dated: 16/05/2017.
- 2.Kharrazi, M., Sencar, H.T., Memon, N.: Blind source camera identification. In: 2004 International Conference on Image Processing, ICIP 2004, vol. 1, pp. 709–712. IEEE (2004)Google Scholar
- 4.Goljan, M., Fridrich, J., Filler, T.: Large scale test of sensor fingerprint camera identification. In: Media Forensics and Security, vol. 7254, p. 72540I. International Society for Optics and Photonics (2009)Google Scholar
- 8.Yang, P., Ni, R., Zhao, Y., Zhao, W.: Source camera identification based on content-adaptive fusion residual networks. Pattern Recogn. Lett. (2017)Google Scholar
- 9.Marra, F., Poggi, G., Sansone, C., Verdoliva, L.: Correlation clustering for PRNU-based blind image source identification. In: IEEE International Workshop on Information Forensics and Security (WIFS), vol. 2016, pp. 1–6. IEEE (2016)Google Scholar
- 11.Lin, X., Li, C.-T.: Large-scale image clustering based on camera fingerprints. IEEE Trans. Inf. Forensics Secur. 12(4), 793–808 (2017)Google Scholar
- 15.Xu, G., Shi, Y.Q.: Camera model identification using local binary patterns. In: 2012 IEEE International Conference on Multimedia and Expo (ICME), pp. 392–397. IEEE (2012)Google Scholar
- 17.Tuama, A., Comby, F., Chaumont, M.: Camera model identification based machine learning approach with high order statistics features. In: 24th European Signal Processing Conference (EUSIPCO), pp. 1183–1187 . IEEE (2016)Google Scholar
- 19.Marra, F., Poggi, G., Sansone, C., Verdoliva, L.: Evaluation of residual-based local features for camera model identification. In: Murino, V., Puppo, E., Sona, D., Cristani, M., Sansone, C. (eds.) ICIAP 2015. LNCS, vol. 9281, pp. 11–18. Springer, Cham (2015). https://doi.org/10.1007/978-3-319-23222-5_2CrossRefGoogle Scholar
- 21.Tuama, A., Comby, F., Chaumont, M.: Camera model identification with the use of deep convolutional neural networks. In: IEEE International Workshop on Information Forensics and Security, 6 p. (2016)Google Scholar
- 23.He, K., Zhang, X., Ren, S., Sun, J.: Deep residual learning for image recognition. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 770–778 (2016)Google Scholar