A multiscale modified minimum spanning forest method for spatial-spectral hyperspectral images classification

EURASIP Journal on Image and Video Processing

Table 5 Information classes, number of training samples, and classification accuracy for the Botswana

	No. of samples	No. of training samples	SVM	SSEPF-RSMSF	MSEPF-RSMSF
Water	270	23	100	99.63	100
Hippo grass	101	23	89.11	100	100
Floodplain grasses1	251	25	100	100	100
Floodplain grasses2	215	23	93.49	100	100
Reeds1	269	23	85.5	96.65	98.88
Riparian	269	23	84.76	100	100
Firescar2	259	23	97.68	100	100
Island interior	203	23	97.04	100	100
Acacia woodlands	314	24	92.04	100	100
Acacia shrublands	248	23	95.56	100	100
Acacia grasslands	305	23	90.49	99.67	99.67
Short mopane	181	23	90.61	100	100
Mixed mopane	268	23	92.16	100	100
Exposed soils	95	23	100	100	100
AA (%)	–	–	93.46	99.7	99.89
OA (%)	–	–	93.23	99.66	99.88
Kappa coefficient	–	–	92.66	99.63	99.87