[GRASS-user] i.gensigset - constraints to generate a sig file

GRASS GRASS user
1

Dear GRASS users,

I am trying to generate a signature file with i.gensigset as an input for i.smap.
So far I am struggling with the creation of the sig file.
Are there any constraints in terms of minimum number of input pixels per class (related to the input image size)?
Is there a relationship between the number of input images in the image group and the maximum classes you can use for training or classification?

Thanks in advance and greetings from Luxembourg,
Christian

Christian Braun

R+D Engineer
Environmental Modelling, Geocomputation

Tel: +352- 425991-6608
Mail: christian.braun@tudor.lu

Resource Centre for Environmental Technologies,
Public Research Centre Henri Tudor,
Technoport Schlassgoart,
66 rue de Luxembourg,
P.O. BOX 144,
L-4002 Esch-sur-Alzette, Luxembourg

2

On Tue, Jan 17, 2012 at 4:03 PM, Christian Braun
<Christian.Braun@tudor.lu> wrote:

Dear GRASS users,

I am trying to generate a signature file with i.gensigset as an input for
i.smap.
So far I am struggling with the creation of the sig file.

What does not work? Please post the error.

Are there any constraints in terms of minimum number of input pixels per
class (related to the input image size)?

I think that a minimum number is needed but I don't know which.

Is there a relationship between the number of input images in the image
group and the maximum classes you can use for training or classification?

Since i.smap is also performing segmentation there may be more
theoretical rather than hard coded limits.

Sidenote: some investigation is ongoing to make i.gensigset faster, hopefully
available soon.

Markus