On Mon, 12 Jul 2004, Peter W. Draper wrote:
> Yes, CCDPACK does this, it's been quite successful at picking out the
> correct arrangement of overlapped images for 10 years now (you probably
> have as much idea as anyone at how strong we can claim "quite successful"
> is). The real secret is that there's a good metric for deciding the most
> likely graph edges, namely the number of paired images. The code is now in
> CVS so anyone can see it at:
>
> http://cvsweb.starlink.ac.uk/cvsweb.cgi/applications/ccdpack
>
> or by using Mark's SCB:
>
> http://andromeda.star.bris.ac.uk/cgi-bin/scb?package=ccdpack
>
> refer them to FINDOFF, but I doubt the code is much use outside of
> CCDPACK.
I'm not sure what language(s) Mr Dobson is used to using, but IMHO
writing this sort of thing in Fortran 77 in this day and age is an
activity best suited to masochists. Working out what the relevant
FINDOFF code was doing and how it worked (admittedly, without much
familiarity with the underlying ideas) made my head hurt for quite
a while.
Mark
--
Mark Taylor Starlink Programmer Physics, Bristol University, UK
[log in to unmask] +44-117-928-8776 http://www.star.bris.ac.uk/~mbt/
|