| (1) |
Color features; gray tone spatial dependence matrices: d = 1; gray gradient co-occurrence matrices: Lg = 10 |
ELM (extreme learning machine): The number of neurons in hidden layer is 80 |
| (2) |
SVMs (support vector machines): Linear kernel function |
| (3) |
SVMs: Linear kernel function; GA (genetic algorithm): population size is 30; crossover rate is 0.7; mutation rate is 0.3; maximum iteration steps is 200 |
| (4) |
kNN (k-nearest neighbor): k = 5, 10, 20 |
| (5) |
Wavelet transformation: Two level wavelet transformation; three kind of wavelets: db1, sym4, haar; all images are resized to be 15 × 30 |
SVMs: Linear kernel function; polynomial kernel function |
| (6) |
LBP (local binary pattern): P = 9; the size of each window is 9 × 9; all images are resized to be 20 × 30 |
| (7) |
Sparse representation: Sample size: 15 × 20, 5 × 10; size of over complete dictionary: 70, 75, 78, 80, 81, 85, 90 |
DE (differential evolution): Population size is 50; crossover rate is 0.7; mutation rate is 0.4; maximum iteration steps is 500 |
| (8) |
Color features; gray tone spatial dependence matrices: d = 1; gray gradient co-occurrence matrices: Lg = 10 sparse representation: size of over complete dictionary: 70, 80, 81, 85 |
