i.cluster

Jerry Thompson thompson at zorro.cecer.army.mil
Tue Apr 7 17:36:24 EDT 1992


> 
>     |I've been asked the following:
>     |
>     |
>     |The i.cluster program requires more than one band ("group")-
>     |If I wanted to try an unsupervised classification on a NDVI
>     |band alone, am I out of luck here?  I suppose I could try 
>     |using AVHRR band 1 or 2 and the NDVI - but I'm not sure if 
>     |that would give proper results.  Any comments?
>     |
>     |Anyone have a handle on this?
> 
> When I wrote i.cluster, I made it require 2 or more bands. That could
> easily be attributed to my novice status as an image processing/
> statistical inference person at the time. There is probably no good
> reason to require 2 or more bands except that the original code
> I was working from would divide by 0 if there was only one band.
> Anyone interested in seeing how the 2 band restriction can be lifted?
> 
> Michael
> 
Why not make a "dummy" file of all value = 1?

Jerry Thompson



More information about the grass-user mailing list