Attacks on Copyright Marking Systems
Proceedings of the Second International Workshop on Information Hiding
A framework for video forensics based on local and temporal fingerprints
ICIP'09 Proceedings of the 16th IEEE international conference on Image processing
Spatial self-synchronizing video watermarking technique
ICIP'09 Proceedings of the 16th IEEE international conference on Image processing
Geometrically invariant watermarking using feature points
IEEE Transactions on Image Processing
Hi-index | 0.00 |
Non-blind watermark detection is still relevant in some applications e.g. traitor tracing. In this case, the auxiliary metadata sent to the detector reveals something about the original content. In this paper, we investigate whether such side-information could also be exploited for registration purpose. To this end, we take a reference non-blind video watermarking system for H.264 AVC CABAC video and show that the watermark auxiliary information could be used in a fingerprint-based registration framework. Our proposed registration strategy operates in two steps: first, identify in the candidate video the best match for each watermarked frame of the master video ; second, discard frames which are most likely to be misregistered (for instance, watermarked frames which have been deleted in the candidate video). In comparison with conventional fingerprint-based techniques, the advantage of this strategy is twofold: (i) the accuracy of the registration naturally adapts to the watermark information density carried by each frame and (ii) it does not require additional storage overhead. Reported experimental results clearly demonstrate that the proposed registration approach largely outperforms other ones relying on independent video fingerprints.