[GRASS-user] field boundary detection with grass gis

[Brad Douglas]

Hello,

I've found that threshold or fuzzy searching (i.zc comes to mind) and
albedo filtering work pretty well on my types of optical imagery without
large computational overhead or models.

Applying a simple cosine (or other) transform on the imagery may also be
helpful in removing light shadows and improve edge detection. YMMV.

I'm also interested in what you've tried, too.

On Tue, 2007-10-02 at 01:32 -0700, Markus Neteler wrote:
> Hello Peter,
> 
> could you tell us which modules you already used?
> 
> I had success with generating synthetic channels with
> r.texture, then running i.smap on those and the original
> data.
> 
> Example for RGB orthophoto classification (with low efforts!):
>  http://mpa.itc.it/markus/ortho_smap/
> Here, we were only interested in a few classes (to investigate
> deer presence for accident mitigation)
> 
> Markus
> 
> 
> Peter Freimuth wrote:
> > 
> > Hallo GRASS User,
> > I am new in GRASS and have to find out for a project what GRASS can do 
> > in regard to field boundary detection based on high resolution satellite 
> > data. I already found some functions for classification and raster to 
> > vector transformation. Can somebody support me with futher info about 
> > additional modules that extend the standard capabilities for better 
> > segmentation results or a nice paper that  describes the best workflow 
> > for such a task in GRASS (6.3 or 6.2).
> > Thanks in advance,
> > Peter


-- 
73, de Brad KB8UYR/6 <rez touchofmadness com>