scanned images

[Trond Rinde]

Hello!

I have scanned a soilmap and imported it into Grass using r.in.tiff, 
i.group, i.points and i.rectify. This worked fine. Thanks for all the
hints from this list. Now I want to modify the map so that each soil-
region get a unique category value. In the scanned map each soil-
region has its own color, or color pattern. However, these scanned colors 
are made up of a composite of pixels with various colors. The different 
pixel values appears in all the soil-regions, only the portions of them 
are different. Therefore, a simple r.reclass will not work. The pixels 
within a soil region must be isolated or grouped together by some 
criteria before they can be assigned new category values. 

I tried to use supervised image classification on the scanned map. This 
did not work because the spectral signature of each of the composite 
colors in the map was not distinct enough.


I guess this is a well known problem to many of the experienced users. 
Any hint would be highly appreciated. Thank you in advance.

Trond Rinde

--
=====================================================================


Siv.ing. Trond Rinde

Institut f. Wasserwirtschaft, Hydrologie und konstruktiven Wasserbau
Universitet f. Bodenkultur
A-1190 Wien, Nussdorfer Laende 11


tel: 	1 36 92 924 352
fax: 	1 36 92 924 350
email:	trinde at donau.boku.ac.at

======================================================================