15040€ — 15042 150°44'E 150*46'E i5Q"48'E 150°S0°E
8.8.9¢
34°10'8
8.01.05
34175
SZLoLE
34148
S ?Ll.v£
34"16'$
$91.r£
450°4WE 150°4ZE — 15044E — 150 49E — TSO'4EE — THRE
sp Bl us RU © ws
NP ;
Figure 4. Results of classification using FS-GA with
SVM classifier.
The improvements of classification accuracy by using FS-GA
technique for different classifiers and combined datasets is
shown in the Figure 5.
Improvements of classification accuracy
by appying GA feature selection techniques
SN
in
3 SVM
M
MANN
8 SOM
Improvements of acuracy (%)
Datasets
Figure 5. Improvements of accuracy by applying FA-GA
techniques for SVM, ANN and SOM classifiers.
The comparison of classification results between the best
classifier in non-FS, FS-GA approach and the multiple classifier
combination using Dempster-Shafer theory (FS-GA-DS) is
given in the table 4 and figure 6.
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B7, 2012
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia
Classifiation accuracy using non-FS, FS-GA and FS-GA-DS methods
©
2
-
©
vds Non-FS
be FS-GA
exiesFS-GA-DS |
Overall classification accuracy
e x
w 4
e
+
59
Figure 6. Comparison of best classification results of non-FS,
FS-GA and FS-GA-DS methods on different datasets.
Although the FS-GA model has already produced significant
increase in the classification accuracy of evaluated multisource
remote sensing datasets, the integration of multiple classifiers
combination with FS-GA method has even further remarkably
improved the classification performance. The FS-GA-DS
algorithm always gave better accuracy than any best single
classifier in all cases. The range of classification improvements
was from 1.24% for the 2"* dataset to 3.07% for the 4™ dataset
as compared to the FS-GA model. Of course, increases in
classification are even much more significant as compared to
the traditional non-FS method. The highest classification
accuracy obtained by the FS-GA-DS model was 88.29% with
the largest combined datasets. The comparison of improvements
in classification performance between FS-GA and non-FS; FS-
GA-DS and FS-GA was given in the figure 7 below.
One of the probably reason behind the successful of the FS-GA-
DS model is its capability to integrate various optimal (or nearly
optimal) solutions given by the GA method for specific
classifier such as SVM, ANN or SOM to enhance the generality
of the final solution.
improvements of classifiation acuracy
# FS-GA vs Non-FS
B FS-GA vs FS-GA-DS
Improvements (26)
Datasets
Figure 7. Improvements of overall classification accuracy
achieved by using FS-GA and FS-GA-DS model.
Overall classification accuracy (%)
Datasets The Majority Voting (MV) algorithm is also very effective for
Non-FS FS-GA FS-GA-DS combining classification results. However, it gave a slightly
1 59.39 61.15 62.56 lower accuracy than the DS algorithm. Results of MV and DS
algorithms were shown in the Table 5.
2 79.97 81.06 82.30
3 81.47 82.37 83.80 Algorithm Datasets
1 1 2 3 4
82.78 85.22 88.29 DS 62.56 82.30 83.80 88.29
Table 4. Comparison of best classification results using non-FS, MV 61.98 82.30 83.52 88.22
FS-GA and FS-GA-DS classifier combination method.
Table 5. Classification accuracy by applying DS and MV
algorithm for classifier combination.
