Optimization Strategy of a Stacked Autoencoder and Deep Belief Network in a Hyperspectral Remote-Sensing Image Classification Model

Table 1

Selection of sample data for ROSIS-3 and HySpex data (number).

	Training samples	Validation samples	Test samples	Total
Asphalt (R)	1345	450	502	2297
Bare ground	831	282	335	1448
Gravel	1193	400	458	2051
Grassland	1002	333	385	1720
Metal sheet	1850	625	762	3237
Brick	952	333	389	1674
Shadows	2247	718	759	3724
Trees	1380	459	510	2349
Water (H)	1148	379	390	1917
Vegetation	1161	388	383	1932
Concrete road	1321	446	449	2216
Magma rock	1007	351	356	1714
Steel	1085	373	377	1835
Glass	633	211	211	1055
Wall	1233	411	411	2055

Note. R is for ROSIS-3 data; H is for HySpex data.