Zhang, X. G., 2000. Introduction to statistical learning theory 
and support vector machines. Journal of Automatization, 26(1), ACKNOWLEDGEMENTS 
pp.32-42. 
This work was supported by Key Laboratory of Geo- 
Kato, Z., and Pong, T.C., 2006. A Markov random field image informatics of State Bureau of Surveying and Mapping (Grant 
segmentation model for color textured images. Image and No. 200727). 
Vision Computing 24, pp. 1103-1114. 
Class 
Water 
bodies 
Road 
Trail 
Shrub and 
grassland 
Agriculture 
Building 
Row 
Total 
User’s 
Accuracy 
(%) 
Water bodies 
108 
0 
0 
0 
0 
0 
108 
100.0 
Road 
0 
134 
0 
2 
4 
0 
140 
95.71 
Trail 
0 
0 
81 
1 
0 
1 
83 
97.59 
Shrub and grassland 
0 
0 
10 
170 
22 
0 
202 
84.16 
Agriculture 
16 
2 
0 
32 
164 
2 
216 
75.93 
Building 
0 
0 
0 
0 
0 
89 
89 
100.0 
Column Total 
124 
136 
91 
205 
190 
92 
838 
Producer’s Accuracy (%) 
87.10 
98.53 
89.01 
82.93 
86.32 
96.74 
Note: Total samples = 838 pixels, correctly classified samples =746 pixels, overall accuracy 89.02%, kappa coefficient k = 0.87. 
Table 1. The confusion matrix of classification based on pixel-based SVM. 
Class 
Water 
bodies 
Road 
Trail 
Shrub and 
grassland 
Agriculture 
Building 
Row 
Total 
User’s 
Accuracy 
(%) 
Water bodies 
124 
0 
0 
0 
0 
0 
124 
100.0 
Road 
0 
136 
0 
0 
0 
0 
136 
100.0 
Trail 
0 
0 
77 
0 
0 
0 
77 
100.0 
Shrub and grassland 
0 
0 
14 
205 
13 
0 
232 
88.36 
Agriculture 
0 
0 
0 
0 
177 
6 
183 
96.72 
Building 
0 
0 
0 
0 
0 
86 
86 
100.0 
Column Total 
124 
136 
91 
205 
190 
92 
838 
Producer’s Accuracy (%) 
100.0 
100.0 
84.62 
100.0 
93.16 
93.48 
Note: Total samples = 838 pixels, correctly classified samples =805 pixels, overall accuracy 96.06%, kappa coefficient k = 0.95. 
Table 2. The confusion matrix of object-oriented classification based on MRF and SVM.