DISTRIBUTED SOURCE CODING AUTHENTICATION

1
DISTRIBUTED SOURCE CODING AUTHENTICATION OF
IMAGES WITH AFFINE WARPING Yao-Chung Lin, David
Varodayan, and Bernd Girod Information Systems
Laboratory, Stanford University, USA
Motivation
Image Authentication Using Distributed Source
Coding
Minimum Decodable Rate
Experiment JPEG2000 compression illegitimate
text banner
Tampered test images
Selected Slepian-Wolf bit rate
or
JPEG2000 compression
Lin, Varodayan, and Girod, ICIP07
Affine Warping Model
Minimum Decodable Rate
Oracle Decoder
Slepian-Wolf Decoder with Expectation Maximization
Receiver Operating Characteristic Curve
Experiment Slightly random affine warping
JPEG2000/JPEG compression illegitimate text
banner 12,000 test images from Barbara, Lena,
Mandrill, and Peppers Using 220 bytes of
authentication data
Original
Rotated by 5 around center
Conclusions
Corresponding Coordinate Estimation

• Novel image authentication using distributed
  source coding combined with Expectation
  Maximization
• Extended robustness against affine warping
  adjustment
• Can authenticate images with affine warping and
  contrast and brightness adjustment (PCS 09)

Tampered
Channel states associatedwith aligned 16x16
blocks