Multimedia Security And ForensicsMultimedia Security And Forensics

Authentication of Digital ImagesAuthentication of Digital Images

Sarah SummersSarah SummersSarah WahlSarah Wahl

CS525 Semester ProjectCS525 Semester ProjectSpring 2006Spring 2006

MotivationMotivationSeeing is believing … or is it?Seeing is believing … or is it?

Easy to be deceivedEasy to be deceived

GoalsGoals

Identify image tampering methods.Identify image tampering methods.

Assess methods available for protecting Assess methods available for protecting images.images.

Assess image authentication techniques.Assess image authentication techniques.

Identify directions for future work.Identify directions for future work.

Categories of Image TamperingCategories of Image Tampering

There are three main categories of image There are three main categories of image tampering:tampering:

EnhancingEnhancing CompositingCompositing Copy/MoveCopy/Move

EnhancingEnhancing

Changing the color Changing the color of objectsof objects

Changing the Changing the weather conditionsweather conditions

Blurring out objectsBlurring out objects

CompositingCompositing

Combining two or Combining two or more images to more images to create a new imagecreate a new image

Copy-MoveCopy-Move

Copying regions of Copying regions of the original image the original image and pasting into and pasting into other areasother areas..

The yellow area has The yellow area has been copied and been copied and moved to conceal moved to conceal the truck.the truck.

What can be done to protect What can be done to protect digital images?digital images?

WatermarkingWatermarkingFragile watermarksFragile watermarksSemi-fragile watermarksSemi-fragile watermarksSelf-embedding watermarksSelf-embedding watermarksDigital cameras with watermarking Digital cameras with watermarking

capabilitiescapabilities

Digital Fingerprinting/SignaturesDigital Fingerprinting/SignaturesDigital cameras with fingerprinting Digital cameras with fingerprinting

capabilitiescapabilities

Digital WatermarkingDigital Watermarking

The basic concept of digital watermarking The basic concept of digital watermarking an image is that a low level signal is an image is that a low level signal is placed directly into the image data. placed directly into the image data.

Any manipulation of the image will impact Any manipulation of the image will impact the watermark and subsequent retrieval of the watermark and subsequent retrieval of the watermark and examination of its the watermark and examination of its condition will indicate if tampering has condition will indicate if tampering has occurred.occurred.

Fragile WatermarksFragile Watermarks

Fragile watermarks are designed to detect Fragile watermarks are designed to detect every possible change in pixel values .every possible change in pixel values .

Variety of Techniques but in most cases, Variety of Techniques but in most cases, the watermark is embedded in the least the watermark is embedded in the least significant bit (LSB) of the image.significant bit (LSB) of the image.

Advantages: Pick up all image Advantages: Pick up all image manipulations – malicious and non-manipulations – malicious and non-maliciousmalicious

Disadvantages: Too sensitiveDisadvantages: Too sensitive

Semi-Fragile WatermarksSemi-Fragile Watermarks They are robust, to a certain extent, and are less They are robust, to a certain extent, and are less

sensitive to pixel modifications. sensitive to pixel modifications.

Techniques:Techniques: Divide image into blocks and utilize bits from each Divide image into blocks and utilize bits from each

block to calculate a spread spectrum noise like signal block to calculate a spread spectrum noise like signal which is combined with DCT coefficients and inserted which is combined with DCT coefficients and inserted as a watermark.as a watermark.

Divide image into blocks, construct watermark in DCT Divide image into blocks, construct watermark in DCT domain from pseudo-random zero-mean unit variance domain from pseudo-random zero-mean unit variance Gaussian numbers, take the inverse DCT and insert Gaussian numbers, take the inverse DCT and insert into the image.into the image.

Advantage: less sensitive than fragile watermarksAdvantage: less sensitive than fragile watermarks

Self-EmbeddingSelf-Embedding

Tampered images result in lost information. The previous Tampered images result in lost information. The previous techniques will only detect and localize areas of interest techniques will only detect and localize areas of interest when authentication is carried out.when authentication is carried out.

Self-embedding allows tamper detection and recovery of Self-embedding allows tamper detection and recovery of missing information.missing information.

General concept is that the image is embedded in itself General concept is that the image is embedded in itself in an encrypted form.in an encrypted form.

Advantage: Potential for original data to be retrieved.Advantage: Potential for original data to be retrieved. Disadvantage: Tampering with the image can remove Disadvantage: Tampering with the image can remove

blocks of the original image making retrieval of content blocks of the original image making retrieval of content impossibleimpossible

Digital Cameras with Digital Cameras with Watermarking CapabilitiesWatermarking Capabilities

Watermarking based on secret key, block ID and Watermarking based on secret key, block ID and content. The image is divided into blocks and content. The image is divided into blocks and each block watermarked using a frequency each block watermarked using a frequency based spread spectrum technique incorporating based spread spectrum technique incorporating the secret key, block ID and block content.the secret key, block ID and block content.

Image of photographers iris is combined with the Image of photographers iris is combined with the camera ID, the hash of the original image and camera ID, the hash of the original image and other details specific to the camera.other details specific to the camera.

Digital Fingerprints/SignaturesDigital Fingerprints/Signatures

Based on the concept of public key Based on the concept of public key encryption.encryption.

Hashed version of image is encrypted Hashed version of image is encrypted using a private key.using a private key.

Encrypted file provides a unique Encrypted file provides a unique signature/fingerprint of the image which signature/fingerprint of the image which can be used to authenticate by decryption can be used to authenticate by decryption with public key.with public key.

Mainly used in transmission of images.Mainly used in transmission of images.

Digital Cameras with Digital Cameras with Fingerprinting CapabilitiesFingerprinting Capabilities

Epson Image Authentication System (IAS)Epson Image Authentication System (IAS) The IAS software in the camera instantly The IAS software in the camera instantly

seals the captured images with an seals the captured images with an invisible digital fingerprint. invisible digital fingerprint.

Verification of image is achieved by any Verification of image is achieved by any PC with Image Authentication System PC with Image Authentication System software installed software installed

Authentication TechniquesAuthentication Techniques

Active AuthenticationActive Authentication Rely on the presence of a watermark or Rely on the presence of a watermark or

fingerprint.fingerprint. Require knowledge original imageRequire knowledge original image Algorithm/key used to embed the watermark Algorithm/key used to embed the watermark

or fingerprint.or fingerprint. Passive AuthenticationPassive Authentication

No requirement of knowledge of original No requirement of knowledge of original image.image.

Does not rely of presence of watermark or Does not rely of presence of watermark or fingerprint.fingerprint.

Passive Authentication Passive Authentication TechniquesTechniques

Detecting Copy-MoveDetecting Copy-Move

Detecting Traces of Re-samplingDetecting Traces of Re-sampling

Detecting Light InconsistenciesDetecting Light Inconsistencies

Copy-Move DetectionCopy-Move Detection

Original Image Tampered Image

Exact Match Robust Match

Copy-Move DetectionCopy-Move Detection

Original Image Tampered Image PCA Detection

Re-sampling DetectionRe-sampling Detection

Original Image Tampered Image

Periodic pattern in FourierTransform of altered region

Fourier Transform of unaltered region

Inconsistencies in LightingInconsistencies in Lighting

Genuine ImageGenuine Image

Tampered ImageTampered Image

Future ResearchFuture Research

Development of a better self embedding Development of a better self embedding technique.technique.

Development of an all inclusive passive Development of an all inclusive passive authentication technique.authentication technique.

ConclusionsConclusions

Digital image forgeries can be used to Digital image forgeries can be used to deceive the public and the authorities.deceive the public and the authorities.

They are here to stay.They are here to stay. Until non destructible/ non removal digital Until non destructible/ non removal digital

watermarks are perfected, passive watermarks are perfected, passive authentication will remain necessary.authentication will remain necessary.

Currently no single passive authentication Currently no single passive authentication technique can detect all types of digital technique can detect all types of digital forgeries.forgeries.

ReferencesReferences Hany Farid, Creating and Detecting Doctored and Virtual Images: Implications to The Child Hany Farid, Creating and Detecting Doctored and Virtual Images: Implications to The Child

Pornography Prevention Act, Pornography Prevention Act, Technical Report, TR2004-518, Dartmouth College, Computer Technical Report, TR2004-518, Dartmouth College, Computer Science.Science.

Detection of Copy-Move Forgery in Digital Images, Jessica Fridrich, David Soukal and Jan Detection of Copy-Move Forgery in Digital Images, Jessica Fridrich, David Soukal and Jan Lukas, Proceedings of Digital Forensic Research Workshop, August 2003, Lukas, Proceedings of Digital Forensic Research Workshop, August 2003, www.ws.binghamton.edu/fridrich/Research/copymove.pdfwww.ws.binghamton.edu/fridrich/Research/copymove.pdf

Detection of image alterations using semi-fragile watermarks, E.T. Lin, C. I. Podilchuk, and Detection of image alterations using semi-fragile watermarks, E.T. Lin, C. I. Podilchuk, and E.J. Delp, E.J. Delp, http://shay.ecn.purdue.edu/~linet/papers/SPIE-2000.pdfhttp://shay.ecn.purdue.edu/~linet/papers/SPIE-2000.pdf

Semi-fragile watermarking for Telltale Tamper Proofing and Authenticating, H. H. Ko and S. J. Semi-fragile watermarking for Telltale Tamper Proofing and Authenticating, H. H. Ko and S. J. Park, Park, http://www.hongik.edu/~sjpark/udt/Semi-Fragile%20Watermarking%20for%20Telltale%20Tahttp://www.hongik.edu/~sjpark/udt/Semi-Fragile%20Watermarking%20for%20Telltale%20Tamper%20Proofing%20and%20A.docmper%20Proofing%20and%20A.doc

Methods for Tamper Detection in Digital Images, Jiri Fridrich, Proc. ACM Workshop on Methods for Tamper Detection in Digital Images, Jiri Fridrich, Proc. ACM Workshop on Multimedia and Security, Orlando, FL, October 30-31, 1999, pp. 19-23, Multimedia and Security, Orlando, FL, October 30-31, 1999, pp. 19-23, http://www.ws.binghamton.edu/fridrich/Research/acm99.dochttp://www.ws.binghamton.edu/fridrich/Research/acm99.doc

Information Authentication for a Slippery New Age, S. Walton, Dr. Dobbs Journal, Vol. 20, No. Information Authentication for a Slippery New Age, S. Walton, Dr. Dobbs Journal, Vol. 20, No. 4, pp 18-26, Apr 19954, pp 18-26, Apr 1995

Blind Detection of Photomontage using Higher Order Statistics, T. Ng, S. Chang and Q. Sun, Blind Detection of Photomontage using Higher Order Statistics, T. Ng, S. Chang and Q. Sun, http://www.ee.columbia.edu/~qibin/papers/qibin2004_iscas_1.pdfhttp://www.ee.columbia.edu/~qibin/papers/qibin2004_iscas_1.pdf

References (continued)References (continued) A Digital Watermark, R. van Schyndel, A. Tirkel and C. OsborneA Digital Watermark, R. van Schyndel, A. Tirkel and C. Osborne , Proceedings of the , Proceedings of the

IEEE International Conference on Image Processing, vol. 2, pp. 86-90, Austin, Texas, IEEE International Conference on Image Processing, vol. 2, pp. 86-90, Austin, Texas, November 1994 November 1994 http://goanna.cs.rmit.edu.au/~ronvs/papers/ICIP94.PDFhttp://goanna.cs.rmit.edu.au/~ronvs/papers/ICIP94.PDF

A Watermark for Image Integrity and Ownership Verification, P. Wong, IS&T’s 1998 Image A Watermark for Image Integrity and Ownership Verification, P. Wong, IS&T’s 1998 Image Processing, Image Quality, Image Capture, Systems Conference, Portland, Oregon, May Processing, Image Quality, Image Capture, Systems Conference, Portland, Oregon, May 1998, pp. 374 – 3791998, pp. 374 – 379

An Invisible Watermarking Technique for Image Verification, M. Yeung and F. Mintzer, An Invisible Watermarking Technique for Image Verification, M. Yeung and F. Mintzer, Proc. ICIP’97, Santa Barbara, California 1997Proc. ICIP’97, Santa Barbara, California 1997

Image watermarking for tamper detection, Jiri Fridrich,Image watermarking for tamper detection, Jiri Fridrich, Proc. ICIP '98, Chicago, Oct 1998, Proc. ICIP '98, Chicago, Oct 1998, http://www.rl.af.mil/programs/shid/downloads/icip98_434.pdfhttp://www.rl.af.mil/programs/shid/downloads/icip98_434.pdf

Methods for Detecting Changes in Digital Images, J. Fridrich, Proc. of The 6th IEEE Methods for Detecting Changes in Digital Images, J. Fridrich, Proc. of The 6th IEEE International Workshop on Intelligent Signal Processing and Communication Systems International Workshop on Intelligent Signal Processing and Communication Systems (ISPACS'98), Melbourne, Australia, 4-6 November 1998, pp. 173–177, (ISPACS'98), Melbourne, Australia, 4-6 November 1998, pp. 173–177, http://www.ws.binghamton.edu/fridrich/Research/ispacs.dochttp://www.ws.binghamton.edu/fridrich/Research/ispacs.doc

A Robust Content Based Digital Signature for Image Authentication, M. Schneider and S. A Robust Content Based Digital Signature for Image Authentication, M. Schneider and S. Chang, Proceedings of the International Conference on Image Processing, 1996, Volume Chang, Proceedings of the International Conference on Image Processing, 1996, Volume 3, Issue , 16-19 Sep 1996 Page(s):227 - 230 3, Issue , 16-19 Sep 1996 Page(s):227 - 230

References (continued)References (continued) A New Fingerprinting Method for Digital Images, V. Fotopoulos and A. N. Skodras, A New Fingerprinting Method for Digital Images, V. Fotopoulos and A. N. Skodras,

http://www.upatras.gr/ieee/skodras/pubs/ans-c35.pdfhttp://www.upatras.gr/ieee/skodras/pubs/ans-c35.pdf

Mehdi Kharrazi, Husrev T. Sencar and Nasir Memon, Blind Source Camera Identification, Mehdi Kharrazi, Husrev T. Sencar and Nasir Memon, Blind Source Camera Identification, International Conference on Image Processing, 2004, ICIP’04, Volume 1, 24-27 Oct. 2004, International Conference on Image Processing, 2004, ICIP’04, Volume 1, 24-27 Oct. 2004, pp. 709 -712pp. 709 -712

Rotation, Scale and Translation Invariant Digital Image Watermarking, J.J.K. O’Ruanaidh Rotation, Scale and Translation Invariant Digital Image Watermarking, J.J.K. O’Ruanaidh and T. Pun, Proceedings of the ICIP, VOl. 1, pp 536-539, Santa Barbara, California, Oct and T. Pun, Proceedings of the ICIP, VOl. 1, pp 536-539, Santa Barbara, California, Oct 1997.1997.

Secure Digital Camera, Paul Blythe and Jessica Fridrich, Secure Digital Camera, Paul Blythe and Jessica Fridrich, http://www.dfrws.org/2004/bios/day3/D3-lyth_Secure_Digital_Camera.pdfhttp://www.dfrws.org/2004/bios/day3/D3-lyth_Secure_Digital_Camera.pdf

Alin C. Popescu and Hany Farid, Exposing Digital Forgeries in Color Filter Array Alin C. Popescu and Hany Farid, Exposing Digital Forgeries in Color Filter Array Interpolated Images, IEEE Transactions on Signal Processing, Vol. 53, Issue 10, Part 2, Interpolated Images, IEEE Transactions on Signal Processing, Vol. 53, Issue 10, Part 2, October 2005, pp 3948-3959October 2005, pp 3948-3959

Epson's Image Authentication for digicams, Epson's Image Authentication for digicams, http://www.dpreview.com/new/9904/99040501epson.asphttp://www.dpreview.com/new/9904/99040501epson.asp

When is Seeing Believing, W. J. Mitchell, Scientific American, pp. 44 -49, February 1994.When is Seeing Believing, W. J. Mitchell, Scientific American, pp. 44 -49, February 1994.

References (continued)References (continued) Exposing digital forgeries by detecting inconsistencies in lighting by M. K. Johnson and H. Farid, ACM Exposing digital forgeries by detecting inconsistencies in lighting by M. K. Johnson and H. Farid, ACM

Multimedia and Security Workshop, New York, NY, 2005, Multimedia and Security Workshop, New York, NY, 2005, http://www.cs.dartmouth.edu/~farid/publications/acm05.pdfhttp://www.cs.dartmouth.edu/~farid/publications/acm05.pdf

Exposing Digital Forgeries by Detecting Traces of Re-sampling, A. C. Popescu and H. Farid, IEEE Exposing Digital Forgeries by Detecting Traces of Re-sampling, A. C. Popescu and H. Farid, IEEE Transactions on Signal Processing, 53(2):758-767, 2005, Transactions on Signal Processing, 53(2):758-767, 2005, http://www.cs.dartmouth.edu/~farid/publications/sp05.pdfhttp://www.cs.dartmouth.edu/~farid/publications/sp05.pdf

Exposing digital forgeries by detecting duplicated image regions, A. C. Popescu and H. Farid, Exposing digital forgeries by detecting duplicated image regions, A. C. Popescu and H. Farid, Technical Report 2004-515, Dartmouth College, http://www.ists.dartmouth.edu/library/tr-2004-515.pdfTechnical Report 2004-515, Dartmouth College, http://www.ists.dartmouth.edu/library/tr-2004-515.pdf

A Tutorial on Principal Components Analaysis, Lindsay I. Smith A Tutorial on Principal Components Analaysis, Lindsay I. Smith http://csnet.otago.ac.nz/cosc453/student_tutorials/principal_components.pdfhttp://csnet.otago.ac.nz/cosc453/student_tutorials/principal_components.pdf

Automatic Estimation of the Projected Light Source Direction, P. Nillius and j. –O. Eklundh, Automatic Estimation of the Projected Light Source Direction, P. Nillius and j. –O. Eklundh, Proceddings of the IEEE Computer Science Conference on Computer Vision and Pattern Recognition, Proceddings of the IEEE Computer Science Conference on Computer Vision and Pattern Recognition, 20012001

Protection of Digital Images Using Self Embedding, J. Fridrich and M. Goljan, Protection of Digital Images Using Self Embedding, J. Fridrich and M. Goljan, Symposium on Content Symposium on Content Security and Data Hiding in Digital MediaSecurity and Data Hiding in Digital Media, New Jersey Institute of Technology, May 14, 1999, , New Jersey Institute of Technology, May 14, 1999, http://www.ws.binghamton.edu/fridrich/Research/nj_may14.dochttp://www.ws.binghamton.edu/fridrich/Research/nj_may14.doc

A Model for Image Splicing, T. Ng and S. Chang, ICIP '04. International Conference on Image A Model for Image Splicing, T. Ng and S. Chang, ICIP '04. International Conference on Image Processing,. Volume 2,  24-27 Oct. 2004 Page(s):1169 - 1172 Vol.2Processing,. Volume 2,  24-27 Oct. 2004 Page(s):1169 - 1172 Vol.2