[GRASS-user] Minimum number of pixels required by i.gensigset (i.smap)

[Nikos Alexandris]

I am trying to find an answer to the question of how many "pixels/per training 
object" are at least required for i.gensigset (and successively i.smap) to 
work. I could not find the answer in the (Bouman and Shapiro, 1994) paper.

Among my notes I have "if the number of pixels (of a training object) is too 
small (<17) they will be ignored...". Can't find the source anymore. I even 
think that this is probably wrong and most likely I saw it in the "i.cluster" 
manual (where the parameter "min_size" defaults to 17).

Any pointers? Thank you, Nikos