Re: [Geotools-gt2-users] Geotools and JAITools

[G. A. <gio...@gm...>]

Thanks Andrea for your reply.
I think that this is an updated package [1] of the processing operations,
where georeferencing is managed in a quite extensive way.
The real doubt (I apologize for my ignorance on this topic) is a more
general point: wether GridCoverage2D instances only keep metadata
informations in memory.

For example, to get a GC from a Geotiff the GeoTiffReader is being used.
What will the GridCoverageReader.read() method read in memory? Only
metadata?

giovanni

[1]
http://svn.osgeo.org/geotools/tags/8.0-M2/modules/library/coverage/src/main/java/org/geotools/coverage/processing/

2011/10/29 Andrea Aime <and...@ge...>

> On Sat, Oct 29, 2011 at 1:33 PM, G. Allegri <gio...@gm...> wrote:
> > Hi Michael, thanks for you reply.
> > I'm reading a lot of documentation to grasp the undergrounds of JAI &
> > friends, and I've reached my real doubt. I have to do some calculations
> (map
> > algebra foundamentally, bot not only this) on big raster thta don't fit
> in
> > memory. I'm trying to figure out the best pattern to streamline
> > read->process->write. I've read that JAI operation chaining can
> > automatically manage this if the source is a tiled Tiff (is it right?),
> but
> > I need to manage this process in the context of Geotools, because I need
> to
> > deal with geographical rasters.
> > So, what I'm looking for are utilities that can help to manage tile based
> > (or any other kind of block) reading, processing and writing.
> > I will go on trying, lost between OOM crashes :)
>
> Ok, so you may need to work on Coverage objects, which wrap a rendered
> image with
> the extra informations to georeference it.
>
> If you look into
> http://svn.osgeo.org/geotools/trunk/modules/unsupported/process-raster/
> there
> are a few processes that take coverages and generate coverages, some
> of them using
> jaitools classes.
> Basically the normal drill is that you leverage the georeferencing
> information to
> setup the grids (e.g, they might be misaligned, using different
> resolutions, so
> you have to compensate all of that using jai affine transforms... and I
> believe
> we don't have examples of this) and then do the computation always using
> jai operators, and finally write out in tiff format.
> If everything along the way is tiled, as well as the output, you should be
> able
> to perform large computations that will leverage JAI tiling and not OOM
>
> Cheers
> Andrea
>
> --
> -------------------------------------------------------
> Ing. Andrea Aime
> GeoSolutions S.A.S.
> Tech lead
>
> Via Poggio alle Viti 1187
> 55054  Massarosa (LU)
> Italy
>
> phone: +39 0584 962313
> fax:      +39 0584 962313
>
> http://www.geo-solutions.it
> http://geo-solutions.blogspot.com/
> http://www.youtube.com/user/GeoSolutionsIT
> http://www.linkedin.com/in/andreaaime
> http://twitter.com/geowolf
>
> -------------------------------------------------------
>