<div>NO NO NO!</div>
<div> </div>
<div>I have 30 CORINE land cover classes to classify.</div>
<div>The number 512, for example, means 5.1.2, is a </div>
<div>nominal number, IS NOT an classe id.</div>
<div> </div>
<div>But may be right: may be the way the classes are </div>
<div>represented in the system cause that problem. I'll test</div>
<div>that..</div>
<div> </div>
<div> </div>
<div>Thanks</div>
<div>JD<br><br></div>
<div class="gmail_quote">On Tue, Apr 6, 2010 at 6:44 PM, Micha Silver <span dir="ltr"><<a href="mailto:micha@arava.co.il">micha@arava.co.il</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="PADDING-LEFT: 1ex; MARGIN: 0px 0px 0px 0.8ex; BORDER-LEFT: #ccc 1px solid">joel dinis wrote:<br><br>
<blockquote class="gmail_quote" style="PADDING-LEFT: 1ex; MARGIN: 0px 0px 0px 0.8ex; BORDER-LEFT: #ccc 1px solid">Hello Micha,<br><br>I put <a href="http://r.info/" target="_blank">r.info</a> <<a href="http://r.info/" target="_blank">http://r.info/</a>> -r amostras
<div class="im"><br>as sugested, get the following:<br>min=112<br>max=512<br><br></div></blockquote>
<div class="im">So, if I understand correctly, you are trying to classify into 400 separate classes?!<br>I'm a beginner with image classification, and others with more experience will know better, but I'll bet that's your problem. There must be many separate training areas (with different cat values) with a very similar spectral signature, and the algorithm is not able to differentiate between them.<br>
<br></div>
<blockquote class="gmail_quote" style="PADDING-LEFT: 1ex; MARGIN: 0px 0px 0px 0.8ex; BORDER-LEFT: #ccc 1px solid">
<div class="im">(This values are index to my classes.)<br><br>JD<br><br></div>
<div class="im">On Tue, Apr 6, 2010 at 4:39 PM, Micha Silver <<a href="mailto:micha@arava.co.il" target="_blank">micha@arava.co.il</a> <mailto:<a href="mailto:micha@arava.co.il" target="_blank">micha@arava.co.il</a>>> wrote:<br>
<br> Nikos Alexandris wrote:<br><br> (sorry for re-sorting the message, just to keep it easy to<br> follow-up)<br><br> joel dinis:<br> <br> > I trying to classify remote sensing imagery, and<br>
> I thought to try the i.smap module, but<br> > when I run the i.gensigset,<br> <br><br> Nikos Alexandris:<br> <br> Please, could you copy-paste the exact command you<br>
used?<br> Does the "maxsig=" (e.g. maxsig=20) make any<br> difference?<br> <br><br> joel dinis:<br><br> <br> In fact, I put maxsig=30.<br>
i.gensigset trainingmap=amostras group=grupo subgroup=subgrupo<br> signaturefile=assinaturas maxsig=30<br> <br><br> What about smaller maxsig's, like:<br><br> - maxsig=25<br>
- maxsig=20<br> - maxsig=15<br><br> ?<br><br> Nikos<br><br> <br> Also it might help to know what is contained in the trainingmap. i.e:<br></div> <a href="http://r.info/" target="_blank">r.info</a> <<a href="http://r.info/" target="_blank">http://r.info</a>> -r amostras
<div>
<div></div>
<div class="h5"><br><br> <br> > I always<br> > get the same problem:<br> > ...<br> > Removed a singular subsignature number 1 (9 remain)<br>
> Removed a singular subsignature number 1 (8 remain)<br> > Removed a singular subsignature number 1 (7 remain)<br> > Removed a singular subsignature number 1 (6 remain)<br>
> Removed a singular subsignature number 1 (5 remain)<br> > Removed a singular subsignature number 1 (4 remain)<br> > Removed a singular subsignature number 1 (3 remain)<br>
> Removed a singular subsignature number 1 (2 remain)<br> > Removed a singular subsignature number 1 (1 remain)<br> > Unreliable clustering. Try a smaller initial<br>
number of<br> clusters<br> > Removed a singular subsignature number 1 (-1 remain)<br> > Unreliable clustering. Try a smaller initial<br> number of<br>
clusters<br> > Number of subclasses is 0<br> > Clustering class 12 (184 pixels)...<br> > Removed a singular subsignature number 1 (9 remain)<br>
> Removed a singular subsignature number 1 (8 remain)<br> > Removed a singular subsignature number 1 (7 remain)<br> > Removed a singular subsignature number 1 (6 remain)<br>
> Removed a singular subsignature number 1 (5 remain)<br> > Removed a singular subsignature number 1 (4 remain)<br> > Removed a singular subsignature number 1 (3 remain)<br>
> Removed a singular subsignature number 1 (2 remain)<br> > Removed a singular subsignature number 1 (1 remain)<br> > Unreliable clustering. Try a smaller initial<br>
number of<br> clusters<br> > Removed a singular subsignature number 1 (-1 remain)<br> > Unreliable clustering. Try a smaller initial<br> number of<br>
clusters<br> > Number of subclasses is 0<br> > i.gensigset complete.<br> ><br> > And afterwards the signature file doesn't have any<br>
> spectral information. I really don't understand whats<br> happening<br> > and the cause to this outcome.<br> ><br> > Can anyone solve this "enigma"?<br>
><br> > Ps: I don't know if this information is important: I<br> collected<br> > about 450 samples distributed by 10 classes. I<br> may also<br>
> add that the average number of pixels per sample<br> is around<br> 50.<br> ><br> > And thanks list, for the previous tips: they were<br>
very<br> useful.<br> <br><br><br><br> _______________________________________________<br> grass-user mailing list<br></div></div> <a href="mailto:grass-user@lists.osgeo.org" target="_blank">grass-user@lists.osgeo.org</a> <mailto:<a href="mailto:grass-user@lists.osgeo.org" target="_blank">grass-user@lists.osgeo.org</a>>
<div class="im"><br> <a href="http://lists.osgeo.org/mailman/listinfo/grass-user" target="_blank">http://lists.osgeo.org/mailman/listinfo/grass-user</a><br><br> This mail was received via Mail-SeCure System.<br>
<br><br> <br><br><br> -- Micha Silver<br> Arava Development Co. +972-52-3665918<br> <a href="http://www.surfaces.co.il/" target="_blank">http://www.surfaces.co.il</a><br><br><br><br><br>This mail was received via Mail-SeCure System.<br>
</div></blockquote>
<div>
<div></div>
<div class="h5"><br><br>-- <br>Micha Silver<br>Arava Development Co. +972-52-3665918<br><a href="http://www.surfaces.co.il/" target="_blank">http://www.surfaces.co.il</a><br><br><br></div></div></blockquote></div><br>