Table 4.

Feature wavelength classification model using ELM

Preprocessing	Training set accuracy (%)					Test set accuracy (%)
	SPA	CARS	IVISSA	CARS-SPA	IVISSA-SPA	SPA	CARS	IVISSA	CARS-SPA	IVISSA-SPA
Raw	92.09	93.87	92.89	94.93	92.53	90.4	92.36	91.2	90.31	91.91
D1st	94.67	96.88	96.71	96.98	97.87	92.53	94.93	95.2	94.93	96
D2nd	95.2	98.13	97.51	94.84	97.51	94.76	97.6	96	93.87	96.53
SG + D1st	94.76	98.31	97.15	97.42	96	92.44	96.8	96.27	94.13	93.42
SG + D2nd	96.53	96.53	96.27	94.93	98.13	91.73	96	95.11	91.02	96
SNV + detrend	93.33	95.2	96.17	93.07	96.8	92.36	93.6	92.27	91.91	94.13