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

EURASIP Journal on Image and Video Processing

Table 1 Information classes, number of training samples, and classification accuracy for the Indian Pines

	No. of samples	No. of training samples	SVM	SSEPF-MMSF	MSEPF-MMSF
Alfalfa	46	15	95.65	100	100
Corn-notill	1428	50	64.85	80.31	86.55
Corn-mintill	830	50	61.57	91.81	90.72
Corn	237	50	86.08	98.73	100
Grass-pasture	483	50	92.55	95.86	94.62
Grass-trees	730	50	87.81	98.36	97.95
Grass-pasture-mowed	28	15	96.43	96.43	100
Hay-windrowed	478	50	99.16	100	100
Oats	20	15	100	100	100
Soybean-notill	972	50	66.46	87.74	88.79
Soybean-mintill	2455	50	47.94	84.95	85.38
Soybean-clean	593	50	71.5	93.57	92.22
Wheat	205	50	99.02	100	100
Woods	1265	50	83.32	96.13	97.15
Buildings-G-T-D	386	50	71.76	99.22	98.96
Stone-steel-towers	93	50	91.4	100	100
AA (%)	–	–	82.21	94.87	95.64
OA (%)	–	–	69.79	90.31	91.45
Kappa coefficient	–	–	66.27	89.18	90.32