Identification of Coffee-Bean Varieties Using Terahertz Detection Technology

Yande Liu; Maopeng Li; Jun Hu; Zhen Xu; Huizhen Cui

doi:10.3788/LOP202158.1630002

Journals >Laser & Optoelectronics Progress >Volume 58 >Issue 16 >Page 1630002 > Article

Laser & Optoelectronics Progress
Vol. 58, Issue 16, 1630002 (2021)

Identification of Coffee-Bean Varieties Using Terahertz Detection Technology

Yande Liu^*, Maopeng Li, Jun Hu, Zhen Xu, and Huizhen Cui

Author Affiliations

School of Mechatronics & Vehicle Engineering, East China Jiaotong University, Nanchang, Jiangxi 330013, China

show less

DOI: 10.3788/LOP202158.1630002 Cite this Article Set citation alerts

Yande Liu, Maopeng Li, Jun Hu, Zhen Xu, Huizhen Cui. Identification of Coffee-Bean Varieties Using Terahertz Detection Technology[J]. Laser & Optoelectronics Progress, 2021, 58(16): 1630002 Copy Citation Text

show less

Fig. 1. Terahertz time domain spectral system schematic

Download full size

Fig. 2. Flow chart of qualitative identification model of three kinds of coffee beans

Download full size

Fig. 3. Mean absorption THz spectra of the three coffee beans

Download full size

Fig. 4. Relationship between number of principal components and variance contribution rate

Download full size

Fig. 5. Correlation between the truth value and the predicted value of the three coffee beans. (a) Rwanda-Yunnan; (b) Kenya-Yunnan; (c) Kenya-Rwanda

Download full size

Fig. 6. SVM prediction set classification results

Download full size

Type of coffee beans	Sample code	Total number of samples	Number of training set samples	Number of sample test sets
Kenyan coffee beans	1	200	150	50
Rwandan coffee beans	2	200	150	50
Yunnan coffee beans	3	200	150	50

Table 1. Classification of training sets and test sets and sample codes of different coffee beans

Object	Accuracy	Entirety
Rwanda	88	92
Yunnan	96	92
Kenya	100	98
Yunnan	96	98
Kenya	100	98
Rwanda	96	98

Table 2. Classification results of PLS-DA model%

Modelingmethod	Pretreatment method	Number of principal components	Kenyan coffee beans		Rwandan coffee beans		Yunnan coffee beans		Overall accuracy /%
Modelingmethod	Pretreatment method	Number of principal components	Number of false positives	Accuracy /%	Number of false positives	Accuracy /%	Number of false positives	Accuracy /%	Overall accuracy /%
SVM	No	9	6	88	0	100	3	94	94
	SG	9	4	92	0	100	2	96	96
	MSC	31	2	96	2	96	5	90	94
	SNV	31	0	100	1	100	2	96	98
	Baseline	7	3	94	0	100	0	100	98
BPNN	No	9	10	80	1	98	4	92	90
	SG	9	11	78	7	86	1	98	87
	MSC	31	13	74	6	88	26	48	70
	SNV	31	17	66	6	88	29	42	65
	Baseline	7	12	76	0	100	6	88	88
RF	No	9	5	90	1	98	6	88	92
	SG	9	4	92	4	92	7	86	90
	MSC	31	2	96	4	92	6	88	92
	SNV	31	5	90	6	88	2	96	91
	Baseline	7	4	92	3	94	7	86	91

Table 3. Comparison of identification results of three modeling algorithms

Yande Liu, Maopeng Li, Jun Hu, Zhen Xu, Huizhen Cui. Identification of Coffee-Bean Varieties Using Terahertz Detection Technology[J]. Laser & Optoelectronics Progress, 2021, 58(16): 1630002

Download Citation

Set citation alerts for the article

Tools

Set citation alerts for the article

Save the article for my favorites

Paper Information