Hyperspectral image classification via contextual deep learning

EURASIP Journal on Image and Video Processing

Table 1 Classification accuracy (%) for the Indian Pines image using 10 % training samples as shown in Fig. 3

Class	Train	Test	SAE-LR	LORSAL-MLL	SOMP	MPM-LBP	CDL-MLR
1	5	49	93.33	100.0	100.0	100.0	100.0
2	143	1291	84.66	87.46	95.05	99.70	99.21
3	83	751	84.39	81.23	96.24	100.0	98.89
4	23	211	73.08	88.41	97.93	100.0	100.0
5	50	447	93.47	97.49	98.68	100.0	99.78
6	75	672	93.41	97.34	98.36	99.73	99.73
7	3	23	100.0	100.0	100.0	100.0	100.0
8	49	440	95.11	97.98	100.0	100.0	100.0
9	2	18	100.0	100.0	100.0	100.0	100.0
10	97	871	85.78	83.64	95.44	98.48	99.89
11	247	2221	83.46	86.31	91.80	98.74	99.22
12	61	553	81.62	91.79	91.71	99.35	98.99
13	21	191	98.52	100.0	100.0	100.0	100.0
14	129	1165	91.77	97.04	98.00	99.92	99.69
15	38	342	81.79	86.18	97.07	98.93	99.43
16	10	85	98.88	100.0	100.0	100.0	100.0
OA			86.85	87.18	92.60	97.50	98.23
AA			89.95	90.83	94.14	98.28	98.70
κ			0.8495	0.8536	0.9155	0.9715	0.9799