their scattering mechanism and hence slight misclassification
was observed in this class in the
RISAT
1 data.
Accuracy Assessment
The overall accuracy of
LISS
-
IV
and
RISAT
-1 data is 91 percent
and 90 percent, respectively. The producer’s accuracy of rice in
LISS
-
IV
and
RISAT
-1 data is 97 percent and 100 percent, respec-
tively, and the user’s accuracy is 97 and 96 percent from
LISS
-IV
and
RISAT
1, respectively; while the rice class kappa values
were 0.94 and 0.92, respectively (Table 2). The producer’s
accuracy of rice class using
RISAT
-1 data is 100 percent because
all rice pixels are classified as rice class.
LISS
-IV imagery clas-
sification is mixing up plantation as well as adjacent rice crop
fields as same class (rice) due to which the producer’s accuracy
is 97.56 percent .Though the producer’s accuracy is 100 per-
cent, the class kappa value may be considered as it takes into
account of both the omission and commission errors (Congal-
ton
et al
., 1991). Due to this fact, the kappa value of rice class
shows lower in
RISAT
-1 than in
LISS
-
IV
. The rice cropped area
Figure 5.
dn
values of Green, Red, and NIR bands of Resourcesat-2,
liss
-
iv
data.
Figure 6.
risat
-1 images of Raney decomposition and classification with ground samples overlaid.
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
July 2015
561
