Ensemble Learning Based Multiple Kernel Principal Component Analysis for Dimensionality Reduction and Classification of Hyperspectral Imagery

Table 3

Overall accuracy, OA (%), and Kappa statistic, , of the Indian Pines data set with different models. The best scores for each class are highlighted in bold face font, and the second best is underlined.

Feature		Raw	PCA	KPCARBF	KPCALap	KPCACau	KPCAHIST	A-KPCA	M-KPCA
# of features		159	2	11	51	31	33	35	20
SVM	C	430.5	279.17	534.67	663.98	663.98	663.98	603.46	346.67
params		2.18	15.22	13.37	13.28	13.27	13.28	16.59	4.74
1		87.04	33.34	88.89	74.07	77.78	90.74	90.74	94.44
2		88.77	40.59	86.82	75.80	82.08	83.19	89.82	89.96
3		86.09	06.95	86.81	65.47	82.61	78.30	88.49	88.60
4		92.31	35.90	91.03	68.38	81.62	82.48	91.45	96.58
5		98.19	58.95	97.79	88.73	95.77	90.74	97.59	98.59
6		98.93	91.43	98.39	97.59	97.99	96.52	99.06	99.33
7		92.31	30.77	92.31	88.46	92.31	84.62	92.31	96.15
8		99.39	95.50	98.98	97.75	98.98	97.34	99.39	99.59
9		90.00	00.00	95.00	90.00	85.00	90.00	95.00	95.00
10		89.67	66.32	91.12	80.58	90.50	88.74	91.22	91.84
11		91.90	73.30	93.27	86.95	91.13	89.55	93.11	94.25
12		94.30	21.01	95.11	80.78	90.88	88.27	95.44	95.60
13		99.06	73.58	99.06	98.11	98.58	97.64	99.06	99.53
14		97.30	86.79	98.07	96.91	97.99	96.52	97.84	97.60
15		80.00	15.53	77.11	53.42	70.26	58.95	82.63	88.95
16		92.63	90.53	94.74	94.74	91.58	90.53	90.53	94.74
OA		92.47	59.78	92.69	83.92	90.16	88.30	93.44	94.28
		0.9130	0.5399	0.9210	0.8215	0.8967	0.8749	0.9317	0.9348