Invisible digital watermarking (IDW) is a type of steganography that aims at concealing information in a medium to prove ownership, integrity or provide additional information.
If it's used to support copyright, its first priority is robustness against destruction and spoofing.
We can define two different needs of copyright: proof of ownership and secure distribution.
In the former, the digital watermark will usually be unique, and it will try to convey an amount of information that suffices in binding the medium with the legal owner.
In the case of distribution, we refer to watermarking as fingerprinting, and the embedded data state a distribution contract between the owner and the receiver. In this way, the watermark will have a variable part to accomodate identity information about the receiver or the issuer of a licence of use, such as a retailer.
If watermarking is used to determine integrity of a piece of data, then its main purpose is to be very weak, so that any minor modification to the medium will destroy it. However, it will have to be highly concealed, too, since an attacker could try to rebuild it after tampering with the data.
These two applications are the ones that gave IDW its name, after the techniques used to protect bills or important documents from tampering or copying.
A less legal-bound application of watermarking is the embedding of additional information in an image. This doesn't need as much robustness as the former two, because the receiver of the medium will usually be interested in the enrichment of the informations. Among the different fields in which it can be useful are:
Additional requests for watermarking algorithms can be ease of encoding or decoding. The former is needed if some retailer has to embed information about the customer on the fly while he is selling the medium. The last is useful if we want to be able to test automatically broadcast or Internet data to detect copyright infringements.
Watermarking techniques have developed enormously in the recent years, as one can see comparing the papers in the '96 and '98 Information Hiding Workshops. The majority of efforts, pushed by money from big firms, has been put on copyright support, since if practical solutions could be found to this problem, then the Internet and new digital channels could be used to distribute data in a quick and inexpensive way.
Another problem is the ease with which a normal user can make very good copies of money bills and other valuable papers using off-the-shelf quite cheap devices, such as scanners and color ink-jet printers. Countries have introduced features that can't be reproduced by these devices in new bills, but since they have to deal with older money, too, they are interested in a way to prevent or discover the authors of such crimes. [GB98]
Compared to classical steganography, digital watermarking adds the problem that usually copyrights last much more than the useful life for secret informations, typically 70 years, so that we can't be satisfied with "computationally infeasible" proofs for resistance to attacks, because we can't know what the capabilities of future attackers will be. [AP97]
We have already provided a description of the various methods for steganography in digital images and other media.
Among them, we can point out as most suitable for invisible digital watermarking:
[Ben96]
[GB98]
[MBR98]
[WW98]
[FBS98]
[AP97]
Previous: Steganography in other media
Next: Attacks
Matteo Fortini