Re: [OSSIM] Thoughts on GRASS and OSSIM

[Mark L. <mlu...@ma...>]

I think we agree with Ned's comments and observations on the two =20
projects.  Several years ago we actually had a GRASS bridge from =20
OSSIM with the libgrass library.  Unfortunately, that hasn't been =20
maintained over time.  The four of us in Melbourne Florida have been =20
able to get government funding over the years that allows us to work =20
on various aspects of the OSSIM baseline.  Our development resources =20
are therefore primarily allocated towards the requirements on those =20
programs.  ImageLinker was basically a prototype evaluation and a =20
means of demonstrating the capabilities of the underlying library.  =20
The osgPlanet library was also built to add functionality to an =20
existing government program.

Currently, we are primarily working fixes, format debugging, and =20
support for web based interfaces and unfortunately don't have a lot =20
of time to dedicate to ImageLinker and documentation ( which needs to =20=

be done).  We are very excited about the prospect of working with =20
Frederic and his group in adding classification and analysis =20
functionality to the baseline.  Perhaps some leverage with GRASS will =20=

make sense with that effort.

Recently we have been in informal collaboration with the guys at =20
Refractions and uDig.  A prototype of uDig with OSSIM layers is in =20
alpha testing, we are hoping to put something together that we can =20
demonstrate at GeoIntel at the end of October.  We have also been =20
talking about adding visual chain editor functionality into uDig and =20
eventually an osgPlanet viewer.  These are goals that we are keeping =20
alive through periodic telecons and development support when we get =20
to it.  uDig certainly has paid a lot of attention to a user friendly =20=

interface and support.

We certainly agree with your comments and goals and I'm always on the =20=

lookout for opportunities to get additional funding on development =20
efforts that can address some of these needs.  In the interim, =20
teaming with other projects and combining resources is probably the =20
best way to proceed.  We are always looking for contributors.  That =20
includes not only development, but documentation, helping with the =20
web site, testing, and networking with other groups.

Mark

On Sep 26, 2005, at 5:17 PM, Ned Horning wrote:

> Greetings,
>
>
> Shortly after the Open Source Geospatial Conference in Minneapolis =20
> there was talk of combining some of the capabilities of GRASS and =20
> OSSIM. Is that discussion continuing? Although I have a limited =20
> knowledge of these two projects it seems like a great idea. =20
> Wouldn=92t it be nice to have the functionality of GRASS, the =20
> performance of OSSIM, and the simple user interface of ImageLinker =20
> (or QGIS for that matter)? =46rom Internet searches I can see there =20=

> was an effort to put together a GRASS/QT library but I haven=92t =20
> noticed any recent work on this project. That also sounds like a =20
> good idea to me and I wonder why it didn=92t get too far.
>
>
> I remain frustrated with the lack of a decent, reasonably full =20
> featured, easy to use, open source image processing package. I=92m =20
> frustrated because it seems like the components are out there but =20
> there is a lack of synergy or something. Is anyone aware of someone =20=

> working on this? The demand is there =96 I hear it all of the time =20
> when we give our training courses.
>
>
> To help put things into perspective here are some of my thoughts on =20=

> GRASS and OSSIM. It would be helpful to hear what other people =20
> think about these two packages and possibly what could/should be =20
> done to improve them to make them more appealing to a broad user =20
> community. This may be a long shot but I would very much like to =20
> start a dialog that would lead to improving existing or creating a =20
> new image processing-focused project.
>
>
> OSSIM (ImageLinker)
>
> OSSIM is a high performance image processing library and =20
> ImageLinker is the GUI application providing access to theOSSIM =20
> libraries. Although significant progress is apparently being made =20
> with the OSSIM libraries there has been little progress on =20
> ImageLinker and associated documentation since the summer of 2004. =20
> ImageLinker can do some task very well such as image visualization =20
> and mosaicing a large number of images but =93simple=94 tasks like =20
> contrast enhancements are cumbersome and image classification is =20
> not supported. A good deal of work needs to go into focusing on =20
> ImageLinker before we could consider using this in our introductory =20=

> remote sensing courses. I expect that much of this would not be too =20=

> difficult but adding other capabilities such as image =20
> classification and vector capabilities would require more time.
>
>
> OSSIM has tremendous potential as a high end image processing =20
> package but the key to success within our target audience is to =20
> package it in a way so the user can easily access the necessary =20
> functionality. The OSSIM community is primarily made up of =20
> developers although with more of a focus on ImageLinker or a =20
> similar application this would become quite popular within the user =20=

> community as well.
>
>
> GRASS
>
> GRASS has, by far, the greatest set of capabilities of the free and =20=

> open source geospatial software I have been following. Although the =20=

> user interface for GRASS has evolved in recent years the underlying =20=

> architecture appears largely unchanged since the original version =20
> was released about 20 years ago and it has become outdated. Adding =20
> a GUI helps but it is not the solution. The two biggest complaints =20
> I hear about GRASS is that it does not run well in the Windows =20
> environment and it is not user friendly or intuitive. The show-=20
> stopper for most people, if they are able to install it, is that =20
> one needs to read a manual simply to learn how to open an image.
>
>
> The GRASS community is active and growing but with some fundamental =20=

> modifications the program could be made much more accessible and =20
> the user community could quickly increase by a factor of 10 and I =20
> view that as a good thing. At this point it is too cumbersome for =20
> us to use in our remote sensing training courses. Although the =20
> functionality is there, most of our students want to use something =20
> that is more intuitive and it needs to run well on Windows and the =20
> other operating systems. Also, the entire =93Database=94 concept in =20=

> GRASS is cumbersome for many people and should be optional. Unless =20
> some changes are made I expect GRASS use will continue to be =20
> limited to modelers and diehard image processing users who don=92t =20
> mind investing the time to learn how to use the program.
>
>
> That=92s enough for now.
>
>
> All the best,
>
>
> Ned
>
> --
>
> Center for Biodiversity and Conservation
>
> American Museum of Natural History
>
> Central Park West @ 79th St
>
> New York, NY 10024
>
> e-mail: ho...@am...
>
> tel: 212-313-7947
>
> fax: 212-769-5292
>
> Home office tel: 802-382-9080
>
> Web site: http://cbc.rs-gis.amnh.org/
>
>
>