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

[mitch_TX]

Hi community,

This semester I'm teaching an Intro Remote Sensing course using commercial
softwares. Since I'm a free spirit, my goal at the end of the semester is to
show the students that everything could be done using GRASS rather than
Envi,Erdas or Idrisi.
Given the fact that they are familiar with GUIs, I'd like to use “graphical”
commands as much as possible (I don't want to scare them)

I have a few questions:

How do I actually save a color composite image as a raster? I know how to
create it and visualize it
(GUI: Add RGB or HIS layer → I choose the bands → Display active layer on
the Map Display)
(Command: d.rgb red=lsat5_1987_40 at landsat green=lsat5_1987_50 at landsat
blue=lsat5_1987_30 at landsat -o)
...but I don't know how to save it, say, to use it as a base image to draw
ground truth for a supervised classification.


The other question is something that has been getting me crazy for a couple
of days: classifications.

(GUI: Imagery → Develop images and groups → Create/edit group. In the 
i.group window I type the name of the group, sub-group and I include rasters
in the group)
(Command: i.group group=lsat_try subgroup=lsat_try
input=lsat5_1987_10 at landsat,lsat5_1987_20 at landsat,lsat5_1987_30 at landsat,lsat5_1987_40 at landsat,lsat5_1987_50 at landsat,lsat5_1987_70 at landsat) 
and the output is:

group [lsat_try] - does not yet exist. Creating...

Adding files to group [lsat_try]

Adding raster map [lsat5_1987_10]

Adding raster map [lsat5_1987_20]

Adding raster map [lsat5_1987_30]

Adding raster map [lsat5_1987_40]

Adding raster map [lsat5_1987_50]

Adding raster map [lsat5_1987_70]

Writing group REF

Done.

Adding files to subgroup [lsat_try]

Adding raster map [lsat5_1987_10]

Adding raster map [lsat5_1987_20]

Adding raster map [lsat5_1987_30]

Adding raster map [lsat5_1987_40]

Adding raster map [lsat5_1987_50]

Adding raster map [lsat5_1987_70]

Writing subgroup REF

Done.


So far so good..right?

Now..
(GUI: Imagery → Classify Image → Clustering input for unsupervised
classification)

In the i.cluster window I type lsat_try for the group, lsat_try for the
sub-group, lsat_sign for the signature file, say 15 classes, 50,50 for the
sampling, lsat_rep for the report.txt (the rest is by default)
(Command: i.cluster group=lsat_try subgroup=lsat_try sigfile=lsat_sign
classes=15 sample=50,50 iterations=30 convergence=98.0 separation=0.0
min_size=17 reportfile=lsat_rep.txt)

output

Reading image ... 



Iteration 1: % Convergence: 80.00 (0s elapsed, 0s left)

Iteration 2: % Convergence: 100.00 (0s elapsed, 0s left)

(Pretty short output, innit?)



Now..
(GUI: Imagery → Classify Image →Maximum likelyhood classification)

I fill the boxes with the required info
(Command: i.maxlik group=lsat_try subgroup=lsat_try sigfile=lsat_sign
class=class_try reject=class_try_rej)

but my output is the following:

Can't read signature file [lsat_sign].


The thing that is literally driving me crazy is that sometimes it works (few
times) others it doesn't (I generally I get error messages saying that the
sub-group doesn't exist in the mapset).


I delete everything and I start over using only the command line:

i.group group=lsat_try sub=lsat_try
in=lsat5_1987_10,lsat5_1987_20,lsat5_1987_30,lsat5_1987_40,lsat5_1987_50,lsat5_1987_70

output:

group [lsat_try] - does not yet exist. Creating...

Adding files to group [lsat_try]

Adding raster map [lsat5_1987_10]

Adding raster map [lsat5_1987_20]

Adding raster map [lsat5_1987_30]

Adding raster map [lsat5_1987_40]

Adding raster map [lsat5_1987_50]

Adding raster map [lsat5_1987_70]

Writing group REF

Done.

Adding files to subgroup [lsat_try]

Adding raster map [lsat5_1987_10]

Adding raster map [lsat5_1987_20]

Adding raster map [lsat5_1987_30]

Adding raster map [lsat5_1987_40]

Adding raster map [lsat5_1987_50]

Adding raster map [lsat5_1987_70]

Writing subgroup REF

Done.


i.cluster group=lsat_try sub=lsat_try sig=lsat_sign classes=15
report=lsat_rep.txt

output
Reading image ... 



Iteration 1: % Convergence: 65.36 (0s elapsed, 0s left)

Iteration 2: % Convergence: 71.92 (0s elapsed, 0s left)

Iteration 3: % Convergence: 86.67 (0s elapsed, 0s left)

Iteration 4: % Convergence: 90.74 (0s elapsed, 0s left)

Iteration 5: % Convergence: 93.32 (0s elapsed, 0s left)

Iteration 6: % Convergence: 95.15 (0s elapsed, 0s left)

Iteration 7: % Convergence: 96.00 (0s elapsed, 0s left)

Iteration 8: % Convergence: 96.29 (0s elapsed, 0s left)

Iteration 9: % Convergence: 96.70 (0s elapsed, 0s left)

Iteration 10: % Convergence: 96.31 (0s elapsed, 0s left)

Iteration 11: % Convergence: 96.70 (0s elapsed, 0s left)

Iteration 12: % Convergence: 96.68 (0s elapsed, 0s left)

Iteration 13: % Convergence: 96.75 (0s elapsed, 0s left)

Iteration 14: % Convergence: 96.84 (0s elapsed, 0s left)

Iteration 15: % Convergence: 96.98 (0s elapsed, 0s left)

Iteration 16: % Convergence: 97.27 (0s elapsed, 0s left)

Iteration 17: % Convergence: 97.54 (0s elapsed, 0s left)

Iteration 18: % Convergence: 97.68 (0s elapsed, 0s left)

Iteration 19: % Convergence: 98.01 (0s elapsed, 0s left)

(Nice output!)



i.maxlik group=lsat_try sub=lsat_try sig=lsat_sign class=lsat_class
rej=lsat_class_rej


output

i.maxlik ... 

(in a fraction of a second while the graphical way – when it works – takes
kind of a minute!!)

I tried to copy the images in the user mapset and start over the GUI way but
no luck!..same error.

Is this  a bug or am I “the bug” :-) ??
I know that if the command line works, you all may not “see what the problem
is”? But as I said I'd like to use the GUI.


Let's try with the Supervised using the group and sub-group that worked
before:
(GUI: Imagery → Classify image → Interactive input for supervised
classifications)

i.class window, say I want to use the NIR band to draw my truths:

 lsat5_1987_40
(good)

group
lsat_try (the group that worked before)
(good) 

sub (if I type list to see the available sub-groups I get: no subgruop
available in the current mapset)

the same happens if I create a group and sub in the user mapset using the
copied images.

What am I doing wrong? :,(



Every input from you all will be more than appreciate.

Thanks,
mitch.

-- 
View this message in context: http://n2.nabble.com/Long-post---Remote-Sensing---Classifications-tp2347807p2347807.html
Sent from the Grass - Users mailing list archive at Nabble.com.