Table 2 Results Obtained by the Proposed NSGA-II and the Standard GA on the three HSI-Pattern data samples

From: Multi-objective band selection algorithm based on NSGA-II for pattern segmentation of textile hyperspectral images

Data Sample

Methods

Trial1

Trial2

Trial3

Mean

Fig. 8(a)

NSGA-II

Number of selected bands

26

   
 

Information Entropy

85,791.2

85,791.2

85,791.2

85,791.2

 

PA(%)

95.40

95.31

95.46

95.39

 

Time(s)

0.696

0.597

0.615

0.636

GA(f1)

Number of selected bands

26

   
 

Information

85,791.2

83,108.2

83,615.8

84,171.7

 

PA(%)

83.16

82.46

82.33

82.65

 

Time(s)

0.043

0.051

0.053

0.049

GA(f2)

Number of selected bands

26

   
 

Information Entropy

78,834.9

78,103.2

78,834.9

78,591.0

 

PA(%)

94.35

91.22

90.81

92.13

 

Time(s)

0.053

0.051

0.041

0.048

Fig. 8(b)

NSGA-II

Number of selected bands

23

   
 

Information Entropy

102,142.6

106,957.1

102,142.6

103,747.4

 

PA(%)

93.74

93.74

93.65

93.71

 

Time(s)

0.552

0.591

0.554

0.566

GA(f1)

Number of selected bands

23

   
 

Information Entropy

107,473.5

107,844.2

108,442.7

107,920.1

 

PA(%)

92.35

92.35

92.35

92.35

 

Time(s)

0.043

0.051

0.032

0.042

GA(f2)

Number of selected bands

23

   
 

Information Entropy

96,921.5

102,732.5

94,253.6

97,969.2

 

PA(%)

92.73

92.73

92.73

92.73

 

Time(s)

0.042

0.047

0.041

0.043

Fig. 8(c)

NSGA-II

Number of selected bands

26

   
 

Information Entropy

84,094.3

84,094.3

84,094.3

84,094.3

 

PA(%)

98.94

98.94

98.94

98.94

 

Time(s)

0.683

0.626

0.578

0.629

GA(f1)

Number of selected bands

26

   
 

Information Entropy

86,700.9

90,902.9

86,700.9

88,101.6

 

Time(s)

0.034

0.046

0.047

0.042

GA(f2)

Number of selected bands

26

   
 

Information Entropy

80,272.1

80,501.2

80,388.7

80,387.3

 

PA(%)

98.94

98.94

98.94

98.94

 

Time(s)

0.029

0.045

0.045

0.040

  1. Bold values denote the optimal performance metrics among the compared methods in the respective evaluation tasks.