[GRASS-user] i.class with i.maxlik

Markus Neteler neteler at osgeo.org
Fri Oct 30 16:43:40 EDT 2009


On Fri, Oct 23, 2009 at 7:07 PM, nawak <nada81 at laposte.net> wrote:
>
> i,
>
> im a french guy so excuse my english.
>
> i try to do a supervised classification with grass but i have an error when
> use i.maxlik
>
> with i.class i create a sigfile. i.maxlik don't find this file...

I can confirm that it gives such a message (6.4.svn).

> howewer the file appear in the subgroup folder

Right.
The problem is that i.maxlik apparently no longer writes the image
statistics. It used to work.

 What I get (North Carolina example):

i.class lsat7_2002_20 group=lsat7_2002 subgroup=lsat7_2002 outsig=testsig

GRASS 6.4.0svn (nc_spm_08):~/grassdata/nc_spm_08/neteler/group/lsat7_2002/subgroup/lsat7_2002/sig/
> cat testsig
#
#water
525

#terrain
897

-> no statistics!

Now this would follow but is useless due to the broken sigfile:
i.maxlik group=lsat7_2002 subgroup=lsat7_2002 sig=ciao class=class
Its error message is unhelpful.

In the i.class ChangeLog I don't see relevant changes, so it must be
a bug in I_write_signatures() perhaps?

Markus


More information about the grass-user mailing list