Table 9.
Recognition accuracy (EER) of recognition systems using a combination of visible-light and thermal images for the recognition problem with PCA (unit: %).
| First SVM Layer Kernel | Accuracy of Recognition Systems Using Combined Visible-Light and Thermal Images | ||||||
|---|---|---|---|---|---|---|---|
| Feature-Level Fusion Approach | Score-Level Fusion Approach | ||||||
| EER | FAR | GAR | Second SVM Layer Kernel | Accuracy | |||
| EER | FAR | GAR | |||||
| Linear | 11.439 | 5.000 | 76.014 | Linear | 11.849 | 5.000 | 74.966 |
| 10.000 | 85.834 | ||||||
| 10.000 | 86.570 | 11.860 | 88.161 | ||||
| 15.000 | 91.274 | ||||||
| 11.440 | 88.561 | 20.000 | 94.270 | ||||
| RBF | 11.863 | 5.000 | 73.472 | ||||
| 15.000 | 92.087 | 10.000 | 85.875 | ||||
| 11.880 | 88.155 | ||||||
| 20.000 | 94.787 | 15.000 | 91.286 | ||||
| 20.000 | 94.490 | ||||||
| RBF | 12.808 | 5.000 | 71.375 | Linear | 11.713 | 5.000 | 73.008 |
| 10.000 | 85.933 | ||||||
| 10.000 | 81.986 | 11.720 | 88.294 | ||||
| 15.000 | 91.483 | ||||||
| 12.820 | 87.205 | 20.000 | 94.297 | ||||
| RBF | 11.753 | 5.000 | 72.040 | ||||
| 15.000 | 89.346 | 10.000 | 85.781 | ||||
| 11.760 | 88.253 | ||||||
| 20.000 | 92.534 | 15.000 | 91.530 | ||||
| 20.000 | 94.368 | ||||||