From: Louis Newstrom (lnewstro@bellsouth.net)
Date: Wed Sep 19 2001 - 08:56:24 MDT
----- Original Message -----
From: "Smigrodzki, Rafal" <SmigrodzkiR@MSX.UPMC.EDU>
> Robert Coyote wrote:
> ### But is it still possible to detect steganographic messages if the
amount
> of information encoded is very small, compared to the amount of
background?
Yes. Codes that use bits of random data have already been devised. They
are harder to crack than codes in which every single bit relates to the
message, but still not impossible.
In this case, the fact that the "random bits" form a picture actually help
identify which bits are the random bits. Even with no information, I could
guess that the most signifigant bits for Red, Green, and Blue values are
random (used for the picture), otherwise the picture would suffer.
Another picture technique, is to use a known picture, and modify the colors
slightly. The difference between the base picture and the one sent would
contain the message. This works well for one-time getting around security.
Sending the same picture more than once is a clear flag that something funny
is going on, and comparing two such pictures would instantly show which bits
are changing and which are static.
> If there are only 100 bytes encoded in a 2 MB photo, I would imagine it
> might be actually impossible to detect the message, without having the
code
> telling you at which particular bits to look.
While this is true, it has nothing to do with the picture technique. It's
just a variation of the old rule that the chance of cracking a code is
proportional to the number of examples. (A one symbol message is always
uncrackable. An infinitely long message is always crackable.)
--- Louis Newstrom lnewstro@bellsouth.net http://personal.lig.bellsouth.net/~newstrom
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 08:10:50 MST