Re: umax-backend

Jonathan A. Buzzard (jab@hex.prestel.co.uk)
Tue, 02 Sep 1997 00:36:44 +0100

R.E.Wolff@BitWizard.nl said:
> If "that is all there is to it", then I don't think this is a valid
> conclusion. I think I'd have to read the whole thesis.

Probably not, but I did not read the thesis myself, I was just assured
the lower two bit where noise. They were actually quite upset about it.

> My explanation of the "uniformity" of the values in the lower bits is
> "there is noise in the source image".

Could be, but it is only one part in 4096, which should be quite easy to
do. I can think of a couple of ways to get some very uniform exposures
on an X-Ray film. Certainly a flood radioactive source is better than
1 in 4096 for uniformity (should be better that 1 in 1 million) if
prepared properly, metal plates are avaliable at high uniformity (an
old hard disc platter would do), you would then be left with the uniformity
of the film (emulsion + backing) which I would hope was better than 1 in
4096.

> What might be a reasonable test is to scan the same image over and
> over again. For every position in the image you can then calculate
> the statistical parameters for that position. The standard deviation
> has something to do with the amount of noise. I expect the pixels on
> sharp edges to show a higher noise figure than other pixels due to
> mechanical positioning problems.

They may well have done that, come to think of it they probably did
(least they used to tie the scanner up for ages). The best uniform image
I can think of, that I have access too is a totally saturated negative
from a 35mm film.

JAB.

-- 
Jonathan A. Buzzard,
34 Kepwell Road, Prudhoe,           Email: jab@hex.prestel.co.uk
Northumberland. NE42 6PD              Tel: +44(0)1661-832195
United Kingdom

--
Source code, list archive, and docs: http://www.azstarnet.com/~axplinux/sane/
To unsubscribe: mail -s unsubscribe sane-devel-request@listserv.azstarnet.com