This is the mail archive of the
gsl-discuss@sources.redhat.com
mailing list for the GSL project.
Re: helping (image processing)
- From: Gerard Jungman <jungman at lanl dot gov>
- To: Jeremy Hayes <cotk_hayes at linuxmail dot org>
- Cc: gsl-discuss at sources dot redhat dot com
- Date: 10 Apr 2003 13:19:06 -0600
- Subject: Re: helping (image processing)
- Organization: Los Alamos National Laboratory
- References: <20030409192438.9738.qmail@linuxmail.org>
On Wed, 2003-04-09 at 13:24, Jeremy Hayes wrote:
> I'm interested in developing Image Analysis functionality for the GSL.
> Do you see this fitting into the GSL framework? If so, how can I help.
>
> Jeremy Hayes
> Computer Science Graduate Research Assistant
> Colorado State University
There are couple of issues with this.
1) Image processing is a huge field, so we would need
a firm plan about what to put in and what to leave out.
2) It will rely on ffts. If you followed the recent discussion
on the mailing list about interfacing to FFTW and the status
of the GSL ffts (starting with a message from Steven Johnson
on Mar 16), then you know that things are a little unsettled.
The ultimate idea, in my opinion, is to have some kind of
compatibility layer so that users can choose which ffts to
use, possibly by choosing at link time in the same way that
a BLAS library is currently chosen at link time.
If the ffts are all invoked in "user space" (explicitly),
then the image processing does not introduce extra problems.
For reasons of flexibility and simplicity, this might be a
good goal for the design. It may not be possible to achieve
this perfectly, and it may not be a problem at all, as long
as the compatibility layer works well.
In any case, we have to solve the problem of creating
a compatibility layer for the ffts. This may be a tedious
way to start the project, depending on your interests, but
I think it is really important at this time. Once something
is in place, then creating a higher layer of spectral
applications should be very pleasant.
What do you think?
Thanks.
--
Gerard Jungman <jungman at lanl dot gov>
Los Alamos National Laboratory