Translator Disclaimer
23 June 2003 Bayesian approach to attack characterization using robust watermarks
Author Affiliations +
Proceedings Volume 5150, Visual Communications and Image Processing 2003; (2003) https://doi.org/10.1117/12.503067
Event: Visual Communications and Image Processing 2003, 2003, Lugano, Switzerland
Abstract
In this paper we propose the use of a Bayesian framework to allow characterisation of image tampering from a library of attacks. We use the double watermarking strategy proposed in our previous work to derive sufficient information to drive the classifier. A non-parametric Bayesian classifier, trained on data derived from Monte Carlo simulations is used. In addition to classification, the effects of varying the input parameters are studied. The results obtained show that the non-parametric Bayesian classifier has a very low misclassification rate for this type of problem. Explanations as to the nature of the results, and some of the practical considerations, are given.
© (2003) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Henry D Knowles, Dominique A. Winne, C. Nishan Canagarajah, and David R. Bull "Bayesian approach to attack characterization using robust watermarks", Proc. SPIE 5150, Visual Communications and Image Processing 2003, (23 June 2003); https://doi.org/10.1117/12.503067
PROCEEDINGS
11 PAGES


SHARE
Advertisement
Advertisement
Back to Top