Sprach Bill Stewart <[EMAIL PROTECTED]>:
> At 08:34 AM 3/22/99 -0600, Brown, R Ken wrote:
> >Probably not news to most of you, but I didn't realise they did this sort of
> >thing.
> >http://www.xerox.com/xsis/dataglph.htm
>
> The basic intent of the system is watermarking, rather than stego in general -
> you can hand somebody a paper copy of your document, marked for them,
> and if you later encounter a Xerox copy, you know whose original it was.
Maybe it's _intended_ for watermarking, but it's very easy to use it
in such a way that you cause someone to reveal more information about
themselves than they wanted. The best (worst?) example I've been able
to come up with is drivers' licenses. It's easy to embed a whole lot
of information on the license using DataGlyphs that the person might
not want to have easily available. It works better than a magnetic
strip because the owner can't make the information unavailable without
visibly defacing the license, and the information will be available
even when the reader doesn't have an MDT handy. Plus, a suitable
imaging device can read the glyphs surreptitiously, even with only a
passing glance at the card.
I've created a basic implementation of DataGlyphs in postscript. The two
features I haven't implemented is the random glyph placement and the
synchronization lattice (which I've found is pretty important for
the part of the reading algorithm that detects where on an image a
dataglyph might be, and at what angle it's printed -- relative to the
scan direction).
My implementation uses Hamming coding for the ECC: 5+1 bits per 24
data bits, to make 30 stroke glyphs.
The Xerox web page describes making the marks 1/100th of an inch long. At
this size, the whole glyph becomes quite easy to pass over, but a careful
examination of the document reveals that there's "something" happening in
the gray. Many people who I've shown it to thought it was a computer
generated maze. :) But the problem with 1/100th of an inch marks is that
the glyph doesn't survive many fax machines: 1/100th of an inch is too close
to the fax machine's resolution. At 1/50th of an inch it survives all the
faxes I've tried, and quite a bit of mutilation to boot. But the glyph area
also becomes immediately noticeable by anyone.
I've found that by changing the width of the strokes, I can change the
apparent gray scale of the mark. Coupling that with a gray scale
bitmap, I can make the glyph area look like some intended image,
especially if the image is larger than the glyph area and I pad to
fill up the space. Many fewer people notice that the area contains
glyphs when I do that.
If anyone's interested in my postscript code, I can probably be convinced
to clean it up and share it. Don't ask to see my scanning code, I'm
ashamed of it. :)
--
Jon Paul Nollmann ne' Darren Senn [EMAIL PROTECTED]
Unsolicited commercial email will be archived at $1/byte/day.
co es agradable que me esten molestando todo el dia cuando yo jamas me
inscribi en su mail-list, cuando trato de que me quiten de su lista, todos
ustedes se la pasan insultandome incluido usted, no quiero que me manden
mas mails, esta manana tenia 194 mails que no me interesan y no solicite
este es mi lugar de trabajo y no estoy jugando como vosotros.
Fabian Casagni, 27/5/1998
