From: Intelligent identification on cotton verticillium wilt based on spectral and image feature fusion
Feature band extraction algorithm | Preprocessing method | SVM classification accuracy (%) | BPNN classification accuracy (%) | ||
---|---|---|---|---|---|
Training set | Test set | Training set | Test set | ||
PCA | Original data | 91.78 | 89.16 | 94.19 | 91.17 |
SG-MSC | 90.58 | 89.36 | 94.19 | 92.37 | |
SG-DT | 87.58 | 86.15 | 92.39 | 91.57 | |
SG-MN | 93.79 | 90.96 | 93.19 | 92.97 | |
SPA | Original data | 94.59 | 92.77 | 94.00 | 91.77 |
SG-MSC | 92.79 | 91.57 | 92.18 | 90.56 | |
SG-DT | 88.38 | 85.74 | 91.18 | 89.96 | |
SG-MN | 95.79 | 93.37 | 95.60 | 93.78 |