Spectral-Spatial Hyperspectral Image Semisupervised Classification by Fusing Maximum Noise Fraction and Adaptive Random Multigraphs

Table 6

Classification results of different techniques for the Pavia University dataset (9 classes).

No.	Pixon-based	PCA-SPCA-2D-SSA	R-VCANet	RN-FSC	iCapsNet	RPNet	MBFSDA	SS-MNF-ARMG
1	45.20	94.83	79.09	87.28	65.84	83.52	86.68	70.83
2	82.00	81.91	71.76	84.33	84.55	54.19	89.22	96.78
3	90.23	85.47	85.45	90.42	29.91	70.70	87.28	96.24
4	74.05	87.99	94.28	78.09	82.70	91.64	79.34	66.42
5	85.13	97.55	99.98	99.56	100.00	100.00	99.33	96.51
6	18.55	71.55	58.44	63.25	40.22	65.30	89.52	100.00
7	92.48	98.77	94.52	52.09	21.40	92.41	88.12	95.49
8	84.17	79.90	73.70	84.81	57.19	74.93	68.88	95.11
9	97.57	96.30	99.82	95.94	0.00	93.24	82.89	77.51
AA	74.38	88.25	84.12	81.94	66.63	80.66	85.70	88.32
OA	69.63	84.50	83.23	81.75	53.54	68.81	86.45	90.32
Kappa	57.72	79.86	67.97	75.84	56.07	61.78	82.31	87.26

Bold values represent the best results among these methods.