[GRASS-user] Apply a segmentation to derive objects

[Nikos Dumakis]

Hi

Regarsding segmentation, has anyone implemented a method to segment an image
in a unsupervised way? (i.smap is supervised).

Or anyone developed some method to do image segmentation?


Thanks
Nikos

On Mon, Jan 18, 2010 at 11:22 PM, Markus Neteler <neteler at osgeo.org> wrote:

> On Mon, Jan 18, 2010 at 6:30 PM, Nikos Dumakis <nikosdu1980 at gmail.com>
> wrote:
> > Greetings all
> > I've been following GRASS-user discussion related with
> pos-classigfication,
> > classification, aggregation and so on. But, while I was readingh those
> > emails I figured that I will need to perform, before classification, an
> > image segmentation to derive objects.
>
> You don't need to perform a segmentation before doing classification but
> i.smap is doing it in a combined way.
>
> > Can anyone give me a clue abvout that because in GRASS book, I only
> founded
> > Image segmentation related with processing large quantities of vectorial
> > data (DEM's).
>
> (DEMs are raster data)
>
> What you need to do is to prepare the statistics with i.gensigset. See
> "radiometric & geometric  supervised" in
>
> http://grass.osgeo.org/wiki/Image_classification
>
> (The book is explaining that as well)
>
> Hope this helps,
> Markus
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.osgeo.org/pipermail/grass-user/attachments/20100302/f84778de/attachment.html