Detecting malicious tampering in digital images Markos Zampoglou - [email protected]Information Technologies Institute (ITI) Centre for Research and Technology Hellas (CERTH) Workshop on Tools for Video Discovery & Verification in Social Media Dec 14, 2017 @ Thessaloniki, Greece
26
Embed
Presentation of the InVID tools for image forensics analysis
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
• Aims:• a comprehensive, self-contained verification toolset
• an evaluation framework for verification tools and protocols
REVEAL Media Verification Assistant (2/2)
Zampoglou, M., Papadopoulos, S., Kompatsiaris, Y., Bouwmeester, R., & Spangenberg, J. (2016, April). Web and Social Media Image Forensics for News Professionals. In SMN@ ICWSM.
• Public demo version• Real-world examples of detections and non-
detections• Tutorials and algorithm explanations• “Magnifying glass” for examining fine details• Scrolling image/map overlay • Export personalized analysis as PDF
France 401Netherlands 262Germany 214UK 181US 153Argentina 96Egypt 52
Current challenges (1/2)
• Absence of automatic methods to analyse metadata• High computational requirements• Interpretation of results not always intuitive
Current challenges (2/2)
• When applied in real-world cases, algorithms perform significantly worse than in research datasets• This could partly be attributed to the effects of Social
Media dissemination
Identifying Untampered Images (1/2)
• Interpreting the results (especially for non-detections) can be an issue
Untampered:
Tampered:
Algorithm: ADQ1 (Lin et al, 2009)
Identifying Untampered Images (2/2)
• Not all algorithms are created equal• Still, some training is usually necessary
Untampered:
Tampered:
Algorithm: ADQ2 (Bianchi et al, 2011)
The pitfalls of image forensics
“MH17 – Forensic Analysis of Satellite Images Released by the Russian Ministry of Defence”
• A report by Bellingcat1 using ELA to prove that the Russian MoD had forged the images
• N. Krawetz: Bellingcat “misinterpreted the results”2
• Bianchi, Tiziano, Alessia De Rosa, and Alessandro Piva. "Improved DCT coefficient analysis for forgery localization in JPEG images." In Acoustics, Speech and Signal Processing (ICASSP), 2011 IEEE International Conference on, pp. 2444-2447. IEEE, 2011.
• Ferrara, Pasquale, Tiziano Bianchi, Alessia De Rosa, and Alessandro Piva. "Image forgery localization via fine-grained analysis of cfa artifacts." Information Forensics and Security, IEEE Transactions on 7, no. 5 (2012): 1566-1577.
• Farid, Hany. "Exposing digital forgeries from JPEG ghosts." Information Forensics and Security, IEEE Transactions on 4, no. 1 (2009): 154-160.
• Fontani, Marco, Tiziano Bianchi, Alessia De Rosa, Alessandro Piva, and Mauro Barni. "A framework for decision fusion in image forensics based on dempster–shafer theory of evidence." Information Forensics and Security, IEEE Transactions on 8, no. 4 (2013): 593-607.
• Lin, Zhouchen, Junfeng He, Xiaoou Tang, and Chi-Keung Tang. "Fast, automatic and fine-grained tampered JPEG image detection via DCT coefficient analysis." Pattern Recognition 42, no. 11 (2009): 2492-2501.
• Mahdian, Babak and Stanislav Saic, “Using noise inconsistencies for blind image forensics,” Image and Vision Computing, vol. 27, no. 10, pp. 1497–1503, 2009.
• Zampoglou, Markos, Symeon Papadopoulos, and Yiannis Kompatsiaris. "Detecting image splicing in the wild (WEB)." In Multimedia & Expo Workshops (ICMEW), 2015 IEEE International Conference on, pp. 1-6. IEEE, 2015.
Evaluation methodology
• Quantitative• Six reference datasets (images + binary masks of tampering
= “ground truth”)• Measures capturing the matching between ground truth
mask and algorithm output• Comparison of 14 algorithms, “best” six plus a newly
proposed one ended up in the tool
• Qualitative• Informal feedback has been received by end users• Pertains to both usability and quality of results
Zampoglou, M., Papadopoulos, S., & Kompatsiaris, Y. (2017). Large-scale evaluation of splicing localization algorithms for web images. Multimedia Tools and Applications, 76(4), 4801-4834.
Quantitative evaluations (1/3)
• To assess the performance of the implemented algorithms, we evaluated their performance on a large number of images
• Benchmark datasets:• Columbia Uncompressed (Hsu and Chang 2006, “COLUMB”), First
Image Forensics Challenge (“CHAL”)1, Fontani et al 2013 (“FON”)
• Additional datasets• The Wild Web Dataset (Zampoglou et al, 2015)