Neural Networks Functions for Public Key Watermarking

Authors:
Justin Picard;Arnaud Robert
Affiliations:
-;-
Venue:
IHW '01 Proceedings of the 4th International Workshop on Information Hiding
Year:
2001

Citing 2
Cited 1

An Asymmetric Public Detection Watermarking Technique

IH '99 Proceedings of the Third International Workshop on Information Hiding
DCT-domain watermarking techniques for still images: detector performance analysis and a new structure

IEEE Transactions on Image Processing

Performance analysis and error exponents of asymmetric watermarking systems

Signal Processing

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper, we propose to use multi-layers neural networks (MLNN) architectures to build public detection functions that approach the detection performance of secret key watermarking. The attractive properties of MLNN for public key watermarking are revealed by a theoretical analysis of linear MLNN. With some approximations, the detection performance of non-linear MLNN, which offer more resistance to attacks, can also be predicted. Experiments on simulated data confirm the theoretical analysis. Also, experiments on real data demonstrate that the selected detection functions are resistant to JPEG compression. Overall, this paper should bring optimism regarding the practical existence of public key watermarking schemes.