Skip to main content

Table 9 Summary of overall classification accuracies for test data, computed with exclusion of unclassified samples

From: Classification of hyperspectral imagery with neural networks: comparison to conventional tools

 

194-band

30-band

13-band

 

%

#uc

%

#uc

%

#uc

ANN

91.89

150

82.13

375

81.24

259

MED

82.04

0

72.85

0

73.29

0

SAM

81.92

145

69.34

186

69.30

156

MLH

  

49.72

0

63.23

0

MHD

  

72.53

0

59.34

0

  1. Numbers under #uc indicate the number of unclassified test samples (out of 4,332 test samples) for each case.