Local and Global Geometric Structure Preserving and Application to Hyperspectral Image Classification

Table 4

Comparison of class accuracy for Indian Pine scene (embedding , %).

Item		Dimension reduction methodologies
Classifier	Class ID	PCA	LapLDA	RP	LPP	LGSPP	JGLDA	LGGSP
5NN	1	83.33	76.09	75.00	82.61	51.35	69.57	91.30
	2	58.44	54.62	57.80	70.24	49.92	70.31	77.24
	3	53.60	38.80	38.85	54.46	36.63	42.89	69.64
	4	41.03	16.88	48.45	46.41	27.59	13.08	85.65
	5	86.92	71.43	82.85	85.30	79.50	74.12	94.20
	6	94.78	65.21	92.35	92.47	87.96	75.89	98.08
	7	92.31	75.00	81.82	82.14	72.73	64.29	96.43
	8	92.64	92.26	89.12	98.33	94.68	99.37	99.79
	9	80.00	50.00	80.00	100.00	80.00	10.00	100.00
	10	77.69	40.74	72.94	51.23	68.16	38.37	82.41
	11	69.49	54.83	63.72	64.73	70.45	65.50	73.16
	12	44.14	26.14	30.98	19.90	21.08	24.45	76.05
	13	96.70	63.90	91.06	99.51	86.49	95.61	99.51
	14	96.45	67.51	90.27	95.65	90.46	94.78	95.18
	15	15.00	21.50	26.81	36.53	24.55	31.09	67.10
	16	85.26	75.27	80.00	90.32	84.42	94.62	94.62
RBF-SVM	1	90.74	76.09	84.38	86.96	64.86	76.09	80.43
	2	81.31	63.45	74.17	77.31	49.77	57.42	82.84
	3	70.86	44.58	57.74	59.40	46.38	57.71	62.29
	4	56.84	12.66	67.53	81.01	59.11	39.24	77.22
	5	95.37	57.56	88.42	89.86	83.11	83.44	92.34
	6	96.12	64.79	92.35	92.60	89.90	77.53	96.85
	7	100.00	53.57	81.82	96.43	81.82	92.86	89.29
	8	94.48	95.82	93.20	99.58	91.44	94.14	99.37
	9	100.00	60.00	100.00	100.00	100.00	75.00	100.00
	10	71.07	34.36	75.23	53.70	64.94	60.70	66.56
	11	79.62	65.34	72.39	80.77	77.42	33.89	84.32
	12	81.92	32.04	51.99	50.93	42.16	60.54	69.31
	13	99.53	69.27	93.30	98.54	76.76	96.59	97.07
	14	94.13	81.50	88.92	95.65	92.23	77.55	94.47
	15	53.95	18.39	46.99	60.62	39.82	72.02	72.28
	16	94.74	77.42	93.33	92.47	83.12	94.62	70.97