Re: Call for stego ideas

From: Arash Partow (arashp_at_hotmail.com)
Date: 07/14/04 

Date: 14 Jul 2004 03:58:41 -0700

Hi,

For images, Wong's watermarking method enables you to have a
high-capacity fragile watermark, now for watermarking thats a
bit useless because the images might undergo compression (ie:
jpeg conversion etc...) However if you are ok with loss-less
image formats, then you can store anywhere from 1-2 bits per
color channel per pixel, so for a 24-bit color image, you can
store up to 6 bits per pixel without degrading the image quality.

Compress + Encrypt the data, in order to avoid detection via
statistical analysis place the data through a LFSR that will
produce a "less-than-random" sequence.

Here is another idea, 2D FFT the image, the grid of frequencies
will be your information base. The logic of encoding is simple,
starting a freq[0,0], if the current bit to be encoded is high
(ie: 1) make the freq magnitude at [0,0] odd, if the bit is low
(ie: 0) make the freq magnitude even.

ie: the freq magnitude at freq[x,y] is 10 and the bit you want
to encode is high (1), what you do is equate freq[x,y] to 9 or
11 depending on what you decide etc..

decoding is simple go through each of the freq in the grid, if
the freq you are currently looking at has an even magnitude then
the bit is 0, if its odd then the bit is 1.

You can increase the potential of storing data (ie: capacity) by
doing an FTT with a tighter frequency resolution.

Some variations of the encoding/decoding technique are prime/no-prime,
multiple of some n / not multiple of some n where n is prime etc...

In both cases you should compress + encrypt + add some FEC to your
data prior to embedding.

I hope this helps...

Arash Partow

__________________________________________________
Be one who knows what they don't know,
Instead of being one who knows not what they don't know,
Thinking they know everything about all things.
http://www.partow.net

Mok-Kong Shen <mok-kong.shen@t-online.de> wrote in message news:<ccp79m$noa$02$1@news.t-online.com>...
> Since stego is comparatively seldom discussed in the group
> than crypto in the narrow sense, I like to request readers
> to contribute their novel ideas for discussion. I would say
> that such ideas are not required initially to satisfy 'any'
> concrete criteria of practical feasibility/efficiency etc.
> If only they have the potential of somehow directly or even
> indirectly supplying inspiration to the eventual development
> (maybe at some later timepoint, as a result of discussion
> here in the group) of something that would be practically
> feasible/efficient etc., then they would be just fine,
> wouldn't they?
>
> A category of stego that I think could be of particular
> interest would be those techniques that are independent of
> any software written specifically for stego purposes, i.e.
> employing only the commonly available scientific/commercial
> software, e.g. the (in a sense) 'standard' graphic,
> text-processing or math software. For there are obviously
> situations of application of stego where the possession of
> stego/crypto software could be a non-trivial risk for the
> persons involved.
>
> Thanks in advance.
>
> M. K. Shen