[5790] in cryptography@c2.net mail archive
Re: grabbed video as a source of entropy
daemon@ATHENA.MIT.EDU (David Honig)
Wed Sep 29 19:57:46 1999
Message-Id: <3.0.5.32.19990927182344.0082d730@pop.sprynet.com>
Date: Mon, 27 Sep 1999 18:23:44 -0700
To: Eugene Leitl <eugene.leitl@lrz.uni-muenchen.de>, <cryptography@c2.net>
From: David Honig <honig@sprynet.com>
In-Reply-To: <14316.11020.11085.750028@lrz.de>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
At 06:53 PM 9/24/99 -0700, Eugene Leitl wrote:
>
>I've recently aquired a video camera (bttv-based 3Com Bigpicture, can
>do 30 fps true color 640x480). I've noticed that under certain
>conditions images can become quite noisy. Does anyone has data on the
>amount and quality of the entropy produced?
Even if I had the same hardware, perhaps the tolerances on my ADCs are
different from yours.
And illumination levels will affect certain kinds of noise.
The point: Measure it.
Use Shannon's entropy formula, for instance.
>I wouldn't mind any pointers to sources extracting, say, LSB from
>grabbed frames.
Distill (irreversibly compress, e.g., by xor'ing several words together)
your data. Measure again; distill again until 1 bit/baud. Hash before use.
Any OTP tool should facilitate or automate these steps.
