Geoscience Reference
In-Depth Information
(a)
(b)
Fig. 12.2
Subset of training set for grass
a
training pixels,
b
reflectance spectrum
reflectance spectrum of the extracted pixels for the respective classes are presented
in Fig.
12.2
b.
pkextract -i 2013_IEEE_GRSS_DF_Contest_FEATURES.tif -s
→
2013_IEEE_GRSS_DF_Contest_Samples_TR_26915.sqlite -f
→
SQLite -otraining_features.sqlite
The second example also creates a training set, but we now use a sample
that consists of multi-polygons. The polygons are obtained from OpenStreetMap
(see also Appendix A.1.1) and have been converted to a Spatialite vector file,
training.sqlite
. We create a training based on the median value for a selec-
tion of polygons. The selection in this example is based on 50% random subset of
Search WWH ::
Custom Search