[WM]: Watermarking requirements

["IQC - Harald Brusewitz" <harald@iqc.se>]

Dear Watermarking collegues,

I want to see a debate on requirements for a useful watermarking system.
I beleive that a rich flow of arguments can improve the direction of research, and hopefully lead to a faster rool-out of useful
products to the benefit of media industry.

First of all I want to see development of a full list of requirements on the wish list, requirements that maybe cannot all be
satisfied at the same time, but still give inspiration to the research community.

Secondly, I want to see our joint efforts to identify what can be fulfilled, or maybe what we think might be possible to fulfill in
the near future.

Third, we need to sort out if current (or near-future) achievements are of sufficient value to the media industry, or otherwise
identify necessary further research.

As a start, I put a few requirements on the wish-list, being sure you will help me to make the list more complete ... or perhaps
more correct!

1. The Watermark shall not by itself destroy or distort the content.
2. Robust to piracy attacks.
   Watermark shall be very difficult to erase without destroying the content.
3. Difficult to detect by outsiders.
   Those who are not certified shall not be able to detect if a watermark is present or not.
4. Easy to detect by insiders.
   Those who are certified shall be able to detect watermark and its "value",
   also after attacks.
5. Detection without original unmarked content.
   It is importnat that it is possible to read the watermark without having access to the
   unmarked content, to minimise the risk that it is spread.
   It is also more practical if such content distribution is not needed.
6. Low computational complexity.
   If CPU power requiremt is low, various copies distributed can have unique fingerprints,
   in order to more easily define who has made pirate copies.

Personally, I think that the second requirement about robustness to attacks is the most interesting, and also most difficult to
satisfy. The pirate community will have enourmous volonteering resources to invent methods to erase watermarks, and they will be
more than happy to distribute such software for free. I can see two strategies to face this
fact:
a. Ignore them. only bother about full quality pirate copies, let the distorted copies free.
b. Win over them. Invent attack-secure watermarking.
Personally, I am afraid that strategy b. will fail.

Now Ladies and Gentlemen, please comment, fill up with missing stuff, and take away redundant bullshit.

- Harald




______________________________________________________________________________

Watermarking Mailing List - http://www.watermarkingworld.org/ml.html
To unsubscribe send email to "majordomo@watermarkingworld.org" with
"unsubscribe watermarking YOURMAIL" in the body.
______________________________________________________________________________