[GRASS-user] Long post - Remote Sensing - Classifications

[mitch_TX]

On Wed, 2009-02-25 at 20:46 -0800, Hamish wrote:
> 
>ok so from the command line 'd.mon x0' should work to open a monitor,
> and you could try the steps first from the command line (keep it simple).
> If all works there you can move to trying launching it from a GUI.
> i.class will always want to run in an xmon + console though.
> 
> this gets i.class working for me using the sample Spearfish imagery
> dataset:
> 
> g.region rast=spot.comp
> d.mon x0
> 

When I do that, I get the following warning using the wxPython GUI.
http://n2.nabble.com/file/n2389455/dx0.png dx0.png 

>d.rast spot.comp
>
> # need to do following in this order to get the r,g,b bands in correct
> #  order in the group/$GROUP/REF file  (??) I'm not really sure.
> 
> # create group
> i.group gr=spotspear in=spot.ms.1,spot.ms.2,spot.ms.3
> 
> # create subgroup
> i.group gr=spotspear subgr=spot.test in=spot.ms.1,spot.ms.2,spot.ms.3
> 

Yes.


# add another map for fun
> i.group gr=spotspear in=spot.comp
> 
# run it
> i.class map=spot.comp group=spotspear1 subgroup=spot.test outsig=test.txt
> 
> 
> it doesn't seem to mind that I was running it from the user1 mapset and
> the images were stored in PERMANENT.
> 
> .... and then work your way though the menus. the result is a signature
> file in $MAPSET/group/$GROUPNAME/subgroup/$SUBGRNAME/sig/sig.txt
> 
> I'm not sure what good results are supposed to look like.
> 

That's the point. Isn't supposed to be a monitor display?
"..divided into three parts. In the upper right corner you see the image. In
the lower right corner, zoomed map portions will be displayed when using the
zoom function. In the left section of the monitor, histograms for the
selected training areas for all channels will be displayed. The training
areas can be digitized...." 
(Neteler, M., H. Mitasova 2008, pp. 320-321)

...right?
It's not happening to me.


> > because I'm runnig this magic on Lenny ;-) (I actually
> > got the 6.4 version from Sid repository).
> 
> nice, I hadn't realized it has been uploaded there yet.
> 
> 
> > is there any way in GRASS to link different map displays
> 
> I am not sure what you mean by that.
> zoom in one moves both?
> 

Yes...


> > and have a "brushing" (update while dragging)
> 
> d.rast.edit?
> 

No. They are general simple windows displaying info regarding coordinates
and Digital Number values like:

Diplay: #
R:# G:# B:#
Proj: UTM ZONE
Map: #, # meters
Lat Long: DD, DD

Moving the cursor on 2 linked displays you have the same geo-info
(you are supposed to link 2 linkable displays) but different RGB values
(generally 2 -- or more -- different bands or color composition are loaded
on different displays).
The meaning of this is allowing a visual inspection (in terms of either
different color composition or shades of gray) of the same area, supported
by a numerical info (DN -- therefore a precise representation of the
reflectance behavior of different surfaces).
I don't know if I made myself clear..I can show you a couple of screen-shots
as soon as I get to the department (it's going to be Envi and Erdas
though...I hope you guys don't mind "commercial" images on this forum :-))



> or even a static d.what.rast?
> 
> I am not sure what you mean by that.
> maybe r.what does what you want?
> 


In order to use r.what you are supposed to use a separate file containing
the points you are querying, right?
No, this is not exactly what I'm talking about...or better, the outcome is
but it's not the interaction that I'm looking for. I'm talking about a
pointer based like d.what.rast one.


> 
> Hamish
> 
> 
> 
>       
> 

mitch



-- 
View this message in context: http://n2.nabble.com/Long-post---Remote-Sensing---Classifications-tp2347807p2389455.html
Sent from the Grass - Users mailing list archive at Nabble.com.