[GRASS5] Training sites and classification

[Markus Neteler]

On Sat, Oct 19, 2002 at 06:39:06PM +0100, Gualter Barbas Baptista wrote:
> How can I define the training sites in GRASS? I tried using i.class but 
> I'm not very clear on how to deal with it.
> In fact I'm a bit confused about the classification processes in GRASS, 
> as the documentation is not very clear to me. Can anyone give some hints 
> on this?

Hi,

did you look at
http://grass.itc.it/gdp/imagery/imagery.ps.gz
or
http://grass.itc.it/gdp/index.html#tutorials
?

Following choices currently exist in GRASS 5 for image classification:
(see above books and tutorials)

               |radiometric, | radiometric,            | radio- and geometric,
               |unsupervised | supervised              | supervised
---------------+-------------+-------------------------+---------------------
Training areas |   --        | a) i.class (in monitor) | a) r.digit
               |             | b) r.digit;             | b) v.digit/v.to.rast
               |             |    v.digit/v.to.rast    |
               |             | c) existing map         | c) existing map
---------------+-------------+-------------------------+---------------------
Statistical    | i.cluster   | a) i.class (in monitor) | i.gensigset
preprocessing  |             | b) i.gensig (from map)  |
---------------+-------------+-------------------------+---------------------
Computation    | i.maxlik    | i.maxlik                | i.smap
---------------+-------------+-------------------------+--------------------- 

Cheers,

 Markus Neteler