Identification of Damage in Pear Using Hyperspectral Imaging Technology

Table 2

Discriminant accuracy of SVM models based on R, A, and K-M spectra.

Spectra	Pretreatment	Accuracy (%), test set/calibration set
		Sound	I	II	III	Total
R	RAW	100/97.92	100/100	87.5/97.92	93.75/100	95.31/98.96
	GF	100/97.92	100/100	87.5/97.92	100/100	96.88/98.96
	SNV	98.44/100	81.25/95.83	81.25/100	81.25/97.92	84.38/98.44
	MSC	93.75/100	81.25/89.58	81.25/93.75	87.50/95.83	85.94/94.79
	COW	87.5/95.83	93.75/83.33	75/87.5	100/93.75	89.06/90.10
	SGD	93.75/95.83	81.25/87.5	68.75/85.42	87.5/95.83	81.25/91.15
A	RAW	100/98.44	100/100	100/98.44	100/100	100/98.98
	GF	100/98.44	93.75/100	93.75/100	100/100	96.88/99.48
	SNV	93.75/100	75/93.75	87.5/95.83	81.25/100	84.38/96.88
	MSC	93.75/100	75/93.75	87.5/95.83	81.25/98.44	84.38/96.88
	COW	100/97.92	87.5/100	87.5/100	93.75/95.83	92.19/98.44
	SGD	100/97.92	62.5/87.5	62.5/91.67	75/89.58	75/91.67
K-M	RAW	100/100	93.75/97.92	87.5/100	93.75/100	93.75/99.48
	GF	100/100	93.75/97.92	87.5/100	93.75/100	93.75/99.48
	SNV	87.5/100	85.71/91.67	68.75/89.58	87.5/95.83	79.69/94.27
	MSC	87.5/100	85.71/91.67	68.75/89.58	87.5/95.83	79.69/94.27
	COW	100/100	87.5/95.83	93.75/100	100/97.92	95.31/98.44
	SGD	75/100	87.5/100	68.75/100	93.75/97.92	79.69/99.48