25
Passive Approaches for Image Forgery Detection Pravin Kakar Primary Adviser: Asst. Prof. Sudha Natarajan, School of Computer Engineering Co-Adviser: Assoc. Prof. Ser Wee, School of Electrical & Electronic Engineering

Passive Approaches for Image Forgery Detection · Estimating the Motion Blur •Yitzhaky et al, Direct Method for Restoration of Motion-blurred Images, Journal of the Optical Society

  • Upload
    others

  • View
    1

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Passive Approaches for Image Forgery Detection · Estimating the Motion Blur •Yitzhaky et al, Direct Method for Restoration of Motion-blurred Images, Journal of the Optical Society

Passive Approaches for Image Forgery Detection

Pravin KakarPrimary Adviser: Asst. Prof. Sudha Natarajan, School of

Computer EngineeringCo-Adviser: Assoc. Prof. Ser Wee, School of Electrical &

Electronic Engineering

Page 2: Passive Approaches for Image Forgery Detection · Estimating the Motion Blur •Yitzhaky et al, Direct Method for Restoration of Motion-blurred Images, Journal of the Optical Society

Outline

• Background & Motivation

• Research Objectives

• Existing Work

• Proposed Method

• Conclusion

Page 3: Passive Approaches for Image Forgery Detection · Estimating the Motion Blur •Yitzhaky et al, Direct Method for Restoration of Motion-blurred Images, Journal of the Optical Society

Background & Motivation

• Tampered images have become pervasive– Counterfeiting– Evidence tampering– Antique Faking– Political Propaganda– Yellow Journalism– Scientific Research– Entertainment– Urban myths

• Forgeries made easy due to– Spread of digital cameras– Easy availability of image manipulation software

Page 4: Passive Approaches for Image Forgery Detection · Estimating the Motion Blur •Yitzhaky et al, Direct Method for Restoration of Motion-blurred Images, Journal of the Optical Society
Page 5: Passive Approaches for Image Forgery Detection · Estimating the Motion Blur •Yitzhaky et al, Direct Method for Restoration of Motion-blurred Images, Journal of the Optical Society

Research Objectives

• Devise new techniques

• Extend existing techniques

• Design efficient algorithms for forgery detection techniques

• Create benchmarking database

Page 6: Passive Approaches for Image Forgery Detection · Estimating the Motion Blur •Yitzhaky et al, Direct Method for Restoration of Motion-blurred Images, Journal of the Optical Society

Forgery Detection Techniques

• Active

– Watermarking

– Signatures

• Passive

– Image Statistics

– Image Content

Page 7: Passive Approaches for Image Forgery Detection · Estimating the Motion Blur •Yitzhaky et al, Direct Method for Restoration of Motion-blurred Images, Journal of the Optical Society

Passive Forgery Detection Techniques

• Farid, H.: A survey of image forgery detection. IEEE Signal Processing Magazine Vol. 2 (2009) 16-25

• Pixel-based Techniques:– Cloning Detection

• Farid et al, Dartmouth College Tech. Rep. TR2004-515 2004• Fridrich et al, Proc. DFRW 2003

– Resampling Detection• Popescu & Farid, IEEE Trans. Signal Processing, 2005

– Splicing Detection• Ng and Chang, Proc. IEEE ICIP, 2004

– Statistical Analyses• Bayram et al, Proc. ESPC, 2005

Page 8: Passive Approaches for Image Forgery Detection · Estimating the Motion Blur •Yitzhaky et al, Direct Method for Restoration of Motion-blurred Images, Journal of the Optical Society

Passive Forgery Detection Techniques

• Format-based Techniques

– JPEG Quantization Table Analysis

• Farid, Dartmouth College Tech. Rep. TR2006-583, 2006

– Double Compression Detection

• Lukas et al, Proc. DFRW, 2003

• Popescu et al, Proc. 6th IWIH, 2004

– JPEG Blocking Detection

• Ye et al, IEEE ICME, 2007

Page 9: Passive Approaches for Image Forgery Detection · Estimating the Motion Blur •Yitzhaky et al, Direct Method for Restoration of Motion-blurred Images, Journal of the Optical Society

Passive Forgery Detection Techniques

• Camera-based Techniques– Chromatic Aberration

• Johnson & Farid, ACM Conf. MM&Sec, 2006

– Color Filter Array Analysis• Popescu & Farid, IEEE Trans. Signal Processing, 2005

– Camera Response• Lin et al, Proc. CVPR, 2005

– Sensor Noise Analysis• Lukas et al, IEEE Trans. Inform. Forensics Sec., 2006

• Gou et al, Proc. IEEE ICIP, 2007

Page 10: Passive Approaches for Image Forgery Detection · Estimating the Motion Blur •Yitzhaky et al, Direct Method for Restoration of Motion-blurred Images, Journal of the Optical Society

Passive Forgery Detection Techniques

• Physics-based Techniques (Johnson & Farid)

– Lighting

• ACM MM&Sec, 2005

• Proc. 9th IWIH, 2007

• IEEE Trans. Inform. Forensics Sec., 2007

– Geometry

• Dartmouth College Tech. Rep. TR2006-579, 2006

• Proc. 6th IWDW, 2007

Page 11: Passive Approaches for Image Forgery Detection · Estimating the Motion Blur •Yitzhaky et al, Direct Method for Restoration of Motion-blurred Images, Journal of the Optical Society
Page 12: Passive Approaches for Image Forgery Detection · Estimating the Motion Blur •Yitzhaky et al, Direct Method for Restoration of Motion-blurred Images, Journal of the Optical Society

Los Angeles Times, April 2003

Page 13: Passive Approaches for Image Forgery Detection · Estimating the Motion Blur •Yitzhaky et al, Direct Method for Restoration of Motion-blurred Images, Journal of the Optical Society

Motion Blur

• Occurs due to

– Camera Shake

– Imaging fast-moving objects

Page 14: Passive Approaches for Image Forgery Detection · Estimating the Motion Blur •Yitzhaky et al, Direct Method for Restoration of Motion-blurred Images, Journal of the Optical Society

Detecting Forgeries based on Blur Inconsistencies

• Hsiao & Pei, “Detecting Digital Tampering by Blur Estimation”, Proceedings of the First International Workshop on Systematic Approaches to Digital Forensic Engineering, 2005.

• Wang et al, “Digital Image Forgery Detection based on the consistency of defocus blur”, International Conference on Intelligent Information Hiding and Multimedia Signal Processing, 2008.

• Sutcu et al, “Tamper Detection based on the regularity of Wavelet Transform Coefficients”, Proceedings of the IEEE International Conference on Image Processing, 2007.

• Zhang & Su, “Detecting Logo-Removal Forgery by Inconsistencies of Blur”, Proceedings of the International Conference on Industrial Mechatronics & Automation”, 2009.

Page 15: Passive Approaches for Image Forgery Detection · Estimating the Motion Blur •Yitzhaky et al, Direct Method for Restoration of Motion-blurred Images, Journal of the Optical Society

Estimating the Motion Blur

• Yitzhaky et al, “Direct Method for Restoration of Motion-blurred Images”, Journal of the Optical Society of America, June 1998.

• Zhang et al, “Estimation of motion parameters from blurred images”, Pattern Recognition Letters (21), 2000.

• Fergus et al, “Removing Camera Shake from a Single Photograph”, ACM SIGGRAPH, 2006.

• Rekleitis, “Optical Flow Recognition from the Power Spectrum of a Single Blurred Image”, Proceedings of the IEEE International Conference on Image Processing, 1996

Page 16: Passive Approaches for Image Forgery Detection · Estimating the Motion Blur •Yitzhaky et al, Direct Method for Restoration of Motion-blurred Images, Journal of the Optical Society

Spectral Matting

• Technique for “soft” segmentation

A. Levin, A. Rav-Acha, and D. Lischinski. Spectral matting. In CVPR, 2007.

Page 17: Passive Approaches for Image Forgery Detection · Estimating the Motion Blur •Yitzhaky et al, Direct Method for Restoration of Motion-blurred Images, Journal of the Optical Society

Blur Estimation from Matting Components

• Gradient of matting components may be used to estimate blur parameters

S. Dai and Y. Wu, “Motion from blur,” in CVPR 2008

Page 18: Passive Approaches for Image Forgery Detection · Estimating the Motion Blur •Yitzhaky et al, Direct Method for Restoration of Motion-blurred Images, Journal of the Optical Society

Proposed Method

Divide image into overlapping blocks

Estimate motion blur for each block

Smooth motion blur estimates

Interpolate motion blur estimates to size of image

Segment image according to motion blur estimates

Accepted for oral presentation at ICME 2010: “Detecting Digital Image Forgeries through Inconsistent Motion Blur”

Page 19: Passive Approaches for Image Forgery Detection · Estimating the Motion Blur •Yitzhaky et al, Direct Method for Restoration of Motion-blurred Images, Journal of the Optical Society

Blur Estimate Measures

Page 20: Passive Approaches for Image Forgery Detection · Estimating the Motion Blur •Yitzhaky et al, Direct Method for Restoration of Motion-blurred Images, Journal of the Optical Society

Detecting Consistent and Inconsistent Regions

Page 21: Passive Approaches for Image Forgery Detection · Estimating the Motion Blur •Yitzhaky et al, Direct Method for Restoration of Motion-blurred Images, Journal of the Optical Society
Page 22: Passive Approaches for Image Forgery Detection · Estimating the Motion Blur •Yitzhaky et al, Direct Method for Restoration of Motion-blurred Images, Journal of the Optical Society

Comparison Results

Page 23: Passive Approaches for Image Forgery Detection · Estimating the Motion Blur •Yitzhaky et al, Direct Method for Restoration of Motion-blurred Images, Journal of the Optical Society

Comparison Results

Page 24: Passive Approaches for Image Forgery Detection · Estimating the Motion Blur •Yitzhaky et al, Direct Method for Restoration of Motion-blurred Images, Journal of the Optical Society

Future Work

• Incorporate improved blur estimation and segmentation techniques

• Extend motion blur-based techniques to video frames.

• Design efficient hardware for execution of the above algorithms

• Research new techniques for detecting digital forgeries

Page 25: Passive Approaches for Image Forgery Detection · Estimating the Motion Blur •Yitzhaky et al, Direct Method for Restoration of Motion-blurred Images, Journal of the Optical Society

Thank you