Panchromatic Image Classification GRASS GIS

Question

How to perform an unsupervised classification of a single panchromatic satellite imagery in GRASS GIS? I tried it many times but I keep being prompted for a second image or other images in the sub-group (band 1-2-3 etc). I am using Landsat pan-chromatic data only only.

Answer 1

For the unsupervised classification of a single panchromatic satellite imagery in GRASS GIS you can use i.segment. It accepts a "group" containing only a single band.

Example (using North Carolina sample dataset: data download + documentation):

GRASS 7.0.1 (nc_spm_08_grass7):~ > i.group group=mono input=lsat7_2002_80
Adding raster map <lsat7_2002_80@PERMANENT> to group

GRASS 7.0.1 (nc_spm_08_grass7):~ > g.region raster=lsat7_2002_80 -p
projection: 99 (Lambert Conformal Conic)
zone:       0
datum:      nad83
ellipsoid:  a=6378137 es=0.006694380022900787
north:      228513
south:      214989.75
west:       629992.5
east:       645012
nsres:      14.25
ewres:      14.25
rows:       949
cols:       1054
cells:      1000246

GRASS 7.0.1 (nc_spm_08_grass7):~ > i.segment group=mono output=lsat7_2002_80_segmented threshold=0.05 minsize=100 goodness=lsat7_2002_80_goodness

Result (original band and two results shown with minsize=5 and minsize=100; goodness of fit maps not shown here):

Note: The colors are arbitrary.