[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
|
[udig-devel] GEF and Raster Support
|
Kenton Williams wrote:
Jody,
Excuse my ignorance, but what is a GEF view? I would be glad to help out testing plugins, but my time is a bit limited this week - I'm attempting to finish up the first draft of my M.S. thesis. I will be glad to help in any way I can though.
Oh finally one I can answer! GEF is the "Graphical Editing Framework".
It is the typical "circles and arrows" move the graphics on the screen
kind of environment. If you start up printing in udig you can see a
basic example where we use it to position content on the page.
The nice part about it is that it really does the Model / View /
Controller split rather well. It is fairly straight forward to hook it
up as a visualization/editing system for any existing model you have
going. If the OSSIM stuff is sufficiently well modeled (or if we can
wrap it on something with a decent model), this would let us easily put
together a palette of operations, and make connections between them etc ...
As far as raster capabilities go, the ones I use most are georeferencing (warping), subsetting, resampling, classification (both supervised and unsupervised), and raster arithmetic. Being able to add, multiply, subtract, divide images by constants
or other images is highly useful. I do really like the OSSIM method of image processing chains rather than something like Idrisi where every step creates a new image file. For this reason, using Idrisi uses a lot of storage space on your hard
drive and requires a lot of cleanup work. As a benefit though, you don't have to worry about your work being lost and you can re-use certain steps.
I too like that approach. Another good example is JAI (but the code is
sufficiently opaque that not many people mange to use it).
I know that georeferencing is a big one, but the Java Advanced Imaging API includes most of what's needed. The only things needed are the polynomial coefficients and the image in a form available to the API. I have been wrestling with free
software intended to perform georeferencing for some time. Hypercube works OK but is a bit of a pain to operate. In fact, my Hypercube processed georeferencing operations have been consistently much more accurate than identical operations in Erdas
Imagine 8.7. I still use Imagine, though, because it's so much easier to use than Hypercube. For those of you that are familiar with Imagine, if its easier to do in a complicated and non-user-friendly program like Imagine, you are in a tight spot!
It seems you already have your JAI mojo happening. I assume you know
geotools is doing a big raster / JAI / processing push right now? Sorry
if this is old news - they are setting up their second breakout meeting
on the subject.
-<http://sourceforge.net/mailarchive/forum.php?thread_id=8000769&forum_id=3008>
They have a bunch of smart guys who right now seem to be more interested
in the lower level JAI ImageIO classes (tile caches and so on). GeoTools
does have a primitive idea of GridCoverage operations backed by JAI
already. In addition to format support they are also interested in
putting together a GridCoverage interface based on some ISO spec (rather
then the moribund OGC they have been following).
I have been planning to write my own open-source Java JAI based georeferencing program utilizing GDAL to fill in this missing link in the Open Source geospatial world. The biggest difficultie comes in getting the images in (GDAL wrapper for Java
isn't available to my knowledge), designing a friendly user interface, and writing an appropriate least-squares adjustment for computation of the polynomial coefficients. All of the difficult stuff is done in JAI. This would, of course, have to
wait 'til I'm done with the M.S. thesis!
Um there are a few Java GDAL wrappers around. The nearest to release
seems to be part of gvSig (a GPL project), I think they have started
taking part in that GC breakout meeting I mentioned above. Chris Holmes
understands the plays involved a bit better then me.
As for the thesis perhaps you would like to attend that IRC meeting and
make sure they are going in a direction you would like to take part in.
Many hands make light work.
I hope this helps, apologies if you knew what was going on already.
Jody
