Blind detection of image manipulation @ PoliMi
1 like1,738 views
The document discusses various techniques for the blind detection of image manipulation without the use of digital watermarks. It outlines pixel-based, format-based, camera-based, physics-based, and geometric-based approaches. It focuses on the use of projective geometry tools and geometric-based techniques like analyzing the assumptions of manual text selection and rectification to known fonts or objects to detect tampering. Key steps involve finding keypoints with SIFT feature detection, matching keypoint pairs with RANSAC, and comparing rectified images to reference samples to judge manipulation.
![Geometric-based Sign manipulations Assumptions: manual selection, [known font]](https://image.slidesharecdn.com/blinddetectionimagemanipulation-110119115241-phpapp02/85/Blind-detection-of-image-manipulation-PoliMi-8-320.jpg)
Blind detection of image manipulation @ PoliMi
- 1. Blind detection of image manipulation Giorgio Sironi, matr. 764852 Politecnico di Milano - Como, January 12th, 2011
- 3. Possible solutions Automated detection of tampering on non-watermarked images Pixel-based Format-based Camera-based Physics-based Geometric-based
- 4. Pixel-based techniques Detection of cloning, resampling , splicing
- 5. Format based techniques Double JPEG , JPEG quantization and blocking
- 6. Camera based techniques Chromatic aberration , sensor noise...
- 7. Physics-based Light direction (2D or 3D)
- 8. Geometric-based Sign manipulations Assumptions: manual selection, [known font]
- 9. What is a photograph? Camera executes a projective transformation ( homography ). We use projective geometry tools.
- 10. Math background Homegeneous coordinates: O is (0,0,1)... Or (0,0,2), or (0,0,3). (10, 15) is (10, 15, 1) or (20, 30, 2)
- 11. Planar homography: 3x3 matrix (linear transformation)
- 12. How do we estimate an homography?
- 13. How do we find the keypoints? SIFT: feature detection
- 14. RANSAC: matching of keypoints pairs
- 15. How do we judge a rectification? Comparison between (binary) rectified image and world sample(s)
- 16. It's an arm race Computer programs support projective geometry tools; do humans?
- 17. References Main Hany Farid, A Survey of Image Forgery Detection (2009), in: IEEE Signal Processing Magazine, 2:26(16-25) Valentina Conotter, Giulia Boato and Hany Farid, Detecting Photo Manipulation on Signs and Billboards, in: International Conference on Image Processing, Hong Kong, 2010 R. Hartley and A. Zisserman, Multiple View Geometry in Computer Vision, Cambridge University Press, 2004. D.G. Lowe, “Distinctive image features from scale-invariant keypoints,” International Journal of Computer Vision, vol. 2, no. 60, pp. 91–110, 2004. Minor Prasad S. and K. R. Ramakrishnan, ON RESAMPLING DETECTION AND ITS APPLICATION TO DETECT IMAGE TAMPERING, in: IEEE International Conference on Multimedia and Expo (ICME 2006),, Jul 09-12, 2006, Toronto, Canada, pp. 1325-1328. Micah K. Johnson and Hany Farid, Exposing Digital Forgeries by Detecting Inconsistencies in Lighting (2005), in: MM&Sec '05 Proceedings of the 7th workshop on Multimedia and security Babak Mahdian and Stanislav Saic, A Bibliography on Blind Methods for Identifying Image Forgery (2010), in: Signal Processing: Image Communication, 25:6(389-399) Sevinc Bayram, Husrev T. Sencar and Nasir Memon, A Survey of Copy-Move Forgery Detection Techniques, in: IEEE Western New York Image Processing Workshop, Rochester, NY, 2008 Nitin Khanna, Aravind K. Mikkilineni, Anthony F. Martone, Gazi N. Ali, George T.-C. Chiu, Jan P. Allebach and Edward J. Delp, A Survey of Forensic Characterization Methods for Physical Devices (2006), in: Digital Investigation, 3:Supplement 1(17-28)
- 18. Thanks for your attention