Table 6 Algorithm settings
From: Potential use of hyperspectral data to classify forest tree species
Radiometric calibration | |
|---|---|
Type | Reflectance |
Output interleave | BSQ |
Output data type | Float |
Scale factor | 1 |
Quick atmospheric correction | |
 Sensor type | AISA |
PCA transformation | |
 Stats X resize factor | 1 |
 Stats Y resize factor | 1 |
 Calculated using | Covariance matrix |
 Output data file | Floating point |
 Select subset from eigenvalues | No |
 Number of output PC bands | 129 |
MNF transformation | |
 Shift difference subset | Full scene |
 Select subset from eigenvalues | No |
 Number of output MNF bands | 129 |
Parallelepiped | |
 Max standard deviation from mean | 3 |
Minimum distance | |
 Max standard deviation from mean | 3 |
 Max distance error | 0 |
Mahalanobis distance | |
 Max standard deviation from mean | 3 |
Maximum likelihood | |
 Max standard deviation from mean | 3 |
 Data scale factor | 255 |
Spectral angle mapping | |
 Maximum angle | 0.1 |
Spectral information divergence | |
 Maximum divergence threshold | 0.05 |
Binary encoding | |
 Minimum encoding threshold | 0 |
Neural networks | |
 Activation | Logistic |
 Training threshold contribution | 0.9 |
 Training rate | 0.2 |
 Training momentum | 0.9 |
 Training RMS exit criteria | 0.1 |
 Number of hidden layers (and nodes) | 1 (129, 36, 3, 7). |
 Number of training iterations | 1000 |
 Minimum output activation threshold | 0 |
Support vector machine | |
 Penalty parameter | 100 |
 Pyramid levels | 0 |
 Classification probability threshold | 0 |