Research Article
A New Scene Classification Method Based on Local Gabor Features
Table 3
Classification performance of different Gabor features and kernel functions on LS dataset.
| Nonlinear kernel | Gabor filter bank | 4 orientations | 8 orientations | 12 orientations |
| RBF | | | | 1 scale | 67.42% | 69.42% | 69.28% | 3 scales | 70.35% | 73.68% | 74.83% | 5 scales | 73.56% | 75.94% | 74.35% | POLY | | | | 1 scale | 64.84% | 69.35% | 68.98% | 3 scales | 71.15% | 76.04% | 74.26% | 5 scales | 72.76% | 75.48% | 75.32% | PMK | | | | 1 scale | 72.38% | 74.29% | 75.25% | 3 scales | 75.54% | 76.46% | 77.65% | 5 scales | 77.42% | 78.85% | 77.84% |
|
|