[GRASS-user] i.maxlik and 'Signature 1 is not valid (ill-conditioned) - ignored'

On Tue, Jul 28, 2009 at 12:18 PM, Thomas Becker<thomasGeo@gmx.de> wrote:

Hi listers,

I am using GRASS64 on Debian Lenny and I am getting warnings about ill-conditioned signatures. As usual the sigfiles are created by i.cluster and I do not know why the first signature is ignored.

I googled the warning but haven't found a solution or workaround for this.

An authentic answer from the author:
http://lists.osgeo.org/pipermail/grass-user/1992-August/014466.html

Perhaps it is originating from a too small cluster?

Markus

Hi Markus and all the others,

the problem is still immanent. It looks like I can create signature files without any ignored or ill-conditioned signatures. I can do so when I increase the min_size attribute as you suggested Markus. Using Quickbird bands for i.cluster I have to set the min_size attribute to about 1000 and can still not avoid ill conditioned signatures in all my images.

However, all the signature files without ill conditioned signatures can not be read by i.maxlik afterwards. I am getting --> ERROR: unable to read signature file <name.sig>.

I will write here as well the command for i.cluster in case I am doing something stupid here.

i.cluster group=T3R.group subgroup=T3R.subgroup sigfile=T3R.sig classes=12 iterations=30 convergence=98.0 separation=0.0 min_size=1000

This command is producing a signature file unable to read by i.maxlik.

Bests Thomas
--
Jetzt kostenlos herunterladen: Internet Explorer 8 und Mozilla Firefox 3 -
sicherer, schneller und einfacher! http://portal.gmx.net/de/go/chbrowser