Journal of Remote Sensing / 2021 / Article / Tab 7

Research Article

Shearlet-Based Structure-Aware Filtering for Hyperspectral and LiDAR Data Classification

Table 7

Classification performance using Raw-H, Raw, 3D-CNN, miniGCN, SAE-LR, NMFL, EPCA, GGF, EMAP, OTVCA and ShearSAF for the Trento dataset with five labeled samples per class as the training set.

ClassRaw-HRaw3D-CNNMiniGCNSAE-LRNMFLEPCAGGFEMAPOTVCAShearSAF

C180.1079.9980.2166.9483.4965.6287.4995.3291.1394.0795.82
C270.5274.0265.5861.7679.9180.8190.7584.7786.2667.0992.43
C396.0096.2588.6686.6272.5184.8082.0996.4298.0986.9489.95
C492.6893.2471.9683.8798.7796.3898.0191.7896.9297.9898.77
C572.1472.5853.3667.5989.6669.2669.4582.1185.1591.2898.23
C656.7360.8755.6076.5062.8479.4576.2873.4779.0166.7068.70
OA78.0179.0964.5373.1087.5680.3983.4586.4389.7689.5793.62
Kappa0.710.730.540.650.840.740.780.820.870.860.91