Study on the Detection and Discrimination of Damaged Jujube Based on Hyperspectral Data

Rui-rui YUAN; Bing WANG; Gui-shan LIU; Jian-guo HE; Guo-ling WAN; Nai-yun FAN; Yue LI; You-rui SUN

doi:10.3964/j.issn.1000-0593(2021)09-2879-07

Spectroscopy and Spectral Analysis, Volume. 41, Issue 9, 2879(2021)

Study on the Detection and Discrimination of Damaged Jujube Based on Hyperspectral Data

Rui-rui YUAN^1、*, Bing WANG^2、*, Gui-shan LIU^{1、1; *;}, Jian-guo HE^1、1;, Guo-ling WAN^1、1;, Nai-yun FAN^1、1;, Yue LI^1、1;, and You-rui SUN^1、1;

Author Affiliations

¹1. School of Food & Wine, Ningxia University, Yinchuan 750021, China

²2. School of Physics and Electronic-Electrical Engineering, Ningxia University, Yinchuan 750021, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(8)

Fig. 1. Damage experimental device of Lingwu long jujube

Download full size

View in Article

Fig. 2. Spectra of Lingwu long jujubes
(a): Original spectra of all samples; (b): Average spectral curves

Download full size

View in Article

Fig. 3. Wavelengths selected by different feature wavelength selection algorithms

Download full size

View in Article

Fig. 4. Stability distribution curve of characteristic variables selected by UVE algorithm

Download full size

View in Article

Table 1. Classification results of PLS-DA of the original and pre-treated spectra

View table

View in Article

Table 1. Classification results of PLS-DA of the original and pre-treated spectra

Pretreatment methods	Principal components	Calibration set(n=270)		Prediction set (n=90)
Pretreatment methods	Principal components	Correct	Accuracy/%	Correct	Accuracy /%
None	16	224	82.96	81	90
SG-1	19	235	87.04	84	93.33
SG-2	17	232	85.93	87	96.67
SNV	18	228	84.44	83	92.22
SNV-SG-1	17	230	85.19	83	92.22
SNV-SG-2	19	246	91.11	87	96.67
Detrending	18	231	85.56	84	93.33
Detrending-SG-1	17	236	87.41	83	92.22
Detrending-SG-2	17	233	86.30	87	96.67

Table 2. Characteristic wavelengths selected by different algorithms

View table

View in Article

Table 2. Characteristic wavelengths selected by different algorithms

Wavelength selection algorithms	Number of wavelengths	Wavelengths/nm
SPA	23	406, 411, 416, 421, 425, 435, 440, 445, 449, 454, 459, 469, 473, 478, 483, 488, 493, 497, 502, 507, 512, 521, 526
IRF	108	406~469 (14), 478~526 (11), 536~641 (23), 651~675 (6), 689~862 (37), 901~944 (10), 963~992 (7)
UVE	68	406, 411, 421, 425, 430, 440, 445, 449, 459, 469, 473, 483, 493, 497, 502, 507, 512, 517, 521, 531, 541, 545, 565, 569, 574, 589, 593, 598, 603, 608, 613, 617, 622, 632, 641, 670, 680, 694, 704, 709, 713, 723, 733, 737, 757, 761, 771, 795, 805, 814, 819, 829, 833, 838, 843, 862, 872, 877, 891, 906, 920, 934, 939, 949, 958, 973, 978, 987
VCPA	13	421, 469, 512, 517, 545, 579, 704, 709, 733, 771, 920, 939, 978
IVISSA	65	421, 425, 430, 435, 440, 445, 449, 454, 483, 488, 493, 497, 502, 507, 512, 517, 521, 526, 531, 560, 565, 569, 574, 579, 584, 589, 593, 598, 603, 608, 613, 617, 622, 627, 632, 637, 670, 675, 704, 709, 713, 733, 757, 761, 766, 771, 848, 853, 857, 862, 901, 906, 910, 915, 920, 925, 930, 934, 939, 968, 973, 978, 982, 987, 992
IRF-SPA	17	445, 517, 541, 617, 689, 733, 737, 757, 776, 781, 829, 838, 862, 915, 925, 939, 982
UVE-SPA	19	473, 502, 517, 541, 603, 608, 622, 641, 733, 737, 795, 805, 843, 862, 891, 920, 949, 973, 987
IVISSA-SPA	15	445, 483, 517, 569, 584, 608, 613, 709, 733, 757, 771, 920, 939, 973, 982

Table 3. The top 10 intervals of feature variables selected by IRF
View table
View in Article
Table 3. The top 10 intervals of feature variables selected by IRF
Ranking Intervals Ranking Intervals
1 6~10 6 74~78
2 7~11 7 80~84
3 17~21 8 5~9
4 19~23 9 8~12
5 20~24 10 10~14

Table 4. The classification results based on characteristic wavelength

View table

View in Article

Table 4. The classification results based on characteristic wavelength

Model	Characteristic wavelength selection method	Number of characteristic wavelengths	PCs	Calibration set (n=270)		Prediction set (n=90)
Model	Characteristic wavelength selection method	Number of characteristic wavelengths	PCs	Correct	Accuracy /%	Correct	Accuracy /%
PLS-DA	SPA	23	19	224	82.96	77	85.56
	IRF	108	17	232	85.93	85	94.44
	UVE	68	16	233	86.30	85	94.44
	VCPA	13	12	204	75.56	67	74.44
	IVISSA	65	17	229	84.81	82	91.11
	IRF-SPA	17	12	204	75.56	67	74.44
	UVE-SPA	19	14	208	77.04	77	85.56
	IVISSA-SPA	15	12	197	72.96	70	77.78
LDA	SPA	23	-	233	86.30	75	83.33
	IRF	108	-	-	-	-	-
	UVE	68	-	-	-	-	-
	VCPA	13	-	211	78.15	60	66.67
	IVISSA	65	-	-	-	-	-
	IRF-SPA	17	-	208	77.04	63	70
	UVE-SPA	19	-	220	81.48	63	70
	IVISSA-SPA	15	-	194	71.85	58	64.44
SVM	SPA	23	-	200	74.07	65	72.22
	IRF	108	-	198	73.33	65	72.22
	UVE	68	-	210	77.78	64	71.11
	VCPA	13	-	166	61.48	47	52.22
	IVISSA	65	-	208	77.04	64	71.11
	IRF-SPA	17	-	171	63.33	45	50
	UVE-SPA	19	-	148	54.81	45	50
	IVISSA-SPA	15	-	112	41.48	31	34.44

Tools

Get Citation

Copy Citation Text

Rui-rui YUAN, Bing WANG, Gui-shan LIU, Jian-guo HE, Guo-ling WAN, Nai-yun FAN, Yue LI, You-rui SUN. Study on the Detection and Discrimination of Damaged Jujube Based on Hyperspectral Data[J]. Spectroscopy and Spectral Analysis, 2021, 41(9): 2879

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites