Maturity Identification of Camellia Seeds Based on Mid- and Far-Infrared Data Fusion

Fig. 1. Camellia fruit and camellia seed sample diagrams. (a) Ripe amellia fruit; (b) camellia seeds after removing the shell of camellia; (c) camellia seed slice sample for mid-infrared spectroscopy test; (d) compressed samples of camellia seeds mixed with polyethylene powder for far-infrared spectroscopy

Download full size

Fig. 2. Mid-infrared absorption spectra of camellia seeds with different oil contents

Download full size

Fig. 3. Far-infrared absorption spectra of camellia seeds with different oil contents

Download full size

Fig. 4. Optimization results of c and g parameters of SVM model based on intermediate data fusion. (a) Three-dimensional view optimized by GS algorithm;(b) contour map optimized by GS algorithm; (c) accuracy curve optimized by GA algorithm;(d) accuracy curve optimized by PSO algorithm

Download full size

Table 1. Oil content of camellia seed
View table
Table 1. Oil content of camellia seed
Picking time Minimum /% Maximum /% Average value /%
September 25th 13.12 18.02 15.23
September 30th 15.29 19.31 17.36
October 6th 18.26 23.50 20.65
October 13th 19.22 23.54 22.03
October 20th 20.60 24.33 23.08

Table 2. Prediction results of SVM model under different feature extraction methods of mid-infrared and far-infrared spectra

View table

Table 2. Prediction results of SVM model under different feature extraction methods of mid-infrared and far-infrared spectra

Spectral type	Feature extraction method	Number of variables	Number of misjudgments in the training set	Number of misjudgments in the prediction set	Training set accuracy /%	Prediction set accuracy /%
Mid-infrared	NONE	1868	14	9	80.00	70.00
	PCA	9	5	6	92.86	80.00
	SPA	32	5	5	92.86	83.33
	UVE	430	7	8	90.00	73.33
Far-infrared	NONE	174	3	4	95.71	86.67
	PCA	9	2	3	97.14	90.00
	SPA	23	2	3	97.14	90.00
	UVE	163	3	4	95.71	86.67

Table 3. Comparison of classification results of optimized SVM model

View table

Table 3. Comparison of classification results of optimized SVM model

Spectral type	Parameter optimization method	SVM parameter		Number of misjudgments	Prediction set accuracy /%
Spectral type	Parameter optimization method	c	g	Number of misjudgments	Prediction set accuracy /%
Mid-infrared	GS	2.2974	6.9644	3	90.00
	GA	2.8523	7.0050	2	93.33
	PSO	5.8397	9.7162	3	90.00
Far-infrared	GS	1.0000	2.8284	2	93.33
	GA	3.3453	0.4628	1	96.67
	PSO	60.1000	0.0100	1	96.67

Table 4. Results of SVM model after parameter optimization based on intermediate data fusion
View table
Table 4. Results of SVM model after parameter optimization based on intermediate data fusion
Optimization algorithm Optimal parameters Number of misjudgments Prediction set accuracy /%
c g
Grid search 0.7579 1.3195 1 96.67
Genetic algorithm 4.3098 1.3823 0 100
Particle swarm algorithm 1.1067 1.2319 0 100

Tools

Get Citation

Copy Citation Text

Xin Ma, Biao Wang, Chun Li, Qingxiao Ma, Yan Teng, Ling Jiang. Maturity Identification of Camellia Seeds Based on Mid- and Far-Infrared Data Fusion[J]. Laser & Optoelectronics Progress, 2022, 59(19): 1930001

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category: Spectroscopy

Received: Aug. 5, 2021

Accepted: Sep. 24, 2021

Published Online: Oct. 12, 2022

The Author Email: Jiang Ling (jiangling@njfu.edu.cn)

DOI:10.3788/LOP202259.1930001

Topics

laser devices and laser physics

Lasers and Laser Optics

Laser physics

laser manufacturing

Instrumentation, Measurement and Metrology

Table 1. Oil content of camellia seed

Table 1. Oil content of camellia seed

Table 2. Prediction results of SVM model under different feature extraction methods of mid-infrared and far-infrared spectra

Table 2. Prediction results of SVM model under different feature extraction methods of mid-infrared and far-infrared spectra

Table 3. Comparison of classification results of optimized SVM model

Table 3. Comparison of classification results of optimized SVM model

Table 4. Results of SVM model after parameter optimization based on intermediate data fusion

Table 4. Results of SVM model after parameter optimization based on intermediate data fusion