Hyperspectral image classification via contextual deep learning

EURASIP Journal on Image and Video Processing

Table 3 Classification accuracy (%) for the University of Pavia image using 9 % training samples as shown in Fig. 9

Class	Train	Test	SAE-LR	LORSAL-MLL	SOMP	MPM-LBP	CDL-MLR
1	597	6034	96.88	100.0	100.0	99.99	100.0
2	1681	16,971	98.30	98.50	97.21	99.00	99.89
3	189	1910	91.09	96.27	92.47	98.81	99.44
4	276	2788	99.15	91.03	74.39	98.24	100.0
5	121	1224	99.85	98.45	93.36	97.20	99.89
6	453	4576	96.44	97.24	99.54	98.61	99.63
7	120	1210	94.12	97.94	96.76	99.18	99.89
8	331	3351	93.27	99.94	99.58	99.90	100.0
9	85	862	100.0	100.0	100.0	97.45	100.0
OA			97.12	98.95	97.84	98.83	99.86
AA			96.57	97.28	93.96	97.36	99.64
κ			0.9615	0.9819	0.9628	0.9799	0.9976