Table 3 Composite datasets and respective training sample used for classification.
Composite no | Datasets | Layer name (Numbers) | Number of training data |
|---|---|---|---|
1 | Different band ratios and spectral Indices (BR) | A total of 15 layers are listed in Table 3 | 150 sample each class × 6 = 900 sample |
2 | OIF derived Sentinel-2 bands (S2_OIF) | A total of 5 layers (S2-Band 2, S2-Band 5, S2-Band 8, S2-Band 11, S2-Band 12, | 50 sample each class × 6 = 300 sample |
3 | BR + S2_OIF | A total of 20 layers | 200 sample each class × 6 = 1200 sample |
4 | Selected optimized layer after applying Stacked Autoencoding for feature extraction (BR + S2_OIF + FE) | A total of 12 layers | 120 sample each class × 6 = 720 sample |
