Category Recognition of Three-Dimensional Fluorescence Spectra of Algae Based on LLE and Gold-SA-SVM

Zhe Liu; Hui Meng; Yongbin Zhang; Weiliang Duan; Ying Chen

doi:10.3788/AOS202242.0430001

Acta Optica Sinica, Volume. 42, Issue 4, 0430001(2022)

Category Recognition of Three-Dimensional Fluorescence Spectra of Algae Based on LLE and Gold-SA-SVM

Zhe Liu, Hui Meng, Yongbin Zhang, Weiliang Duan, and Ying Chen^*

Key Laboratory of Test/Measurement Technology and Instrument of Hebei Province, School of Electrical Engineering, Yanshan University, Qinhuangdao, Hebei 066004, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(9)

Fig. 1. Flowchart of Gold-SA-SVM classification model

Download full size

Fig. 2. 3D fluorescence spectra and contour maps of algae. (a) 3D fluorescence spectrum of Synechococcus elongatus; (b) contour map of Synechococcus elongatus; (c) 3D fluorescence spectrum of Chlorella; (d) contour map of Chlorella; (e) 3D fluorescence spectrum of Aureococcus anophagefferens; (f) contour map of Aureococcus anophagefferens

Download full size

Fig. 3. 3D fluorescence spectra of Chlorella at different stages of growth cycle. (a) 1st day; (b) 3rd day; (c) 6th day; (d) 9th day; (e) 12th day; (f) 15th day

Download full size

Fig. 4. Change of accuracy of LLE-SVM for different K and d

Download full size

Fig. 5. Fluorescence spectrum characteristics of Aureococcus anophagefferens, Chlorella, and Synechococcus elongatus. (a) Aureococcus anophagefferens; (b) Chlorella; (c) Synechococcus elongatus

Download full size

Table 1. Feature values extraced from LLE algorithm

View table

Table 1. Feature values extraced from LLE algorithm

Feature data	PC1	PC2	PC3	PC4	PC5
Sample 1	-1.00587	-2.54722	1.43672	0.32511	0.04593
Sample 2	-0.95251	-2.06625	0.91802	0.05146	-0.03223
Sample 3	-1.00584	-2.54735	1.44139	0.37201	0.08087
︙	︙	︙	︙	︙	︙
Sample 270	2.82157	-0.71266	0.50778	-2.63531	0.37356

Table 2. Classification accuracy of different dimension reduction methods%
View table
Table 2. Classification accuracy of different dimension reduction methods%
Dimension reduction method LDA PCA LLE
Accuracy 95.1 95.6 96.3

Table 3. Confusion matrices of different classification models

View table

Table 3. Confusion matrices of different classification models

Model	Actual	Predict
Model	Actual	Synechococcuselongatus	Chlorella	Aureococcusanophagefferens
	Synechococcus elongatus	90	0	0
NNLS in Ref. [6]	Chlorella	6	73	11
	Aureococcus anophagefferens	6	0	84
	Synechococcus elongatus	88	1	1
Bayesian in Ref. [7]	Chlorella	4	80	6
	Aureococcus anophagefferens	1	0	89
	Synechococcus elongatus	90	0	0
GA-SVM inRefs. [13-14]	Chlorella	3	84	3
	Aureococcus anophagefferens	1	0	89
	Synechococcus elongatus	90	0	0
PSO-SVM inRefs. [15-16]	Chlorella	1	87	2
	Aureococcus anophagefferens	1	1	88
	Synechococcus elongatus	90	0	0
Gold-SA-SVMin this paper	Chlorella	0	90	0
	Aureococcus anophagefferens	0	0	90

Table 4. Comparison of evaluation indexes of different classification models%

View table

Table 4. Comparison of evaluation indexes of different classification models%

Model	Synechococcuselongatus			Chlorella			Aureococcusanophagefferens			A
Model	P	R	F₁	P	R	F₁	P	R	F₁	A
NNLS in Ref. [6]	88.2	100.0	93.8	100.0	81.1	89.6	88.4	93.3	90.8	91.5
Bayesian in Ref. [7]	94.6	98.8	92.7	97.8	88.9	98.9	96.2	93.6	95.7	95.2
GA-SVM in Refs. [13-14]	95.7	100.0	97.8	100.0	93.3	96.6	96.7	98.9	97.8	97.4
PSO-SVM in Refs. [15-16]	97.8	100.0	98.9	98.9	96.7	97.8	97.8	87.8	97.8	98.1
Gold-SA-SVM in this paper	100.0	100.0	100.0	100.0	100.0	100.0	100.0	100.0	100.0	100.0

Tools

Get Citation

Copy Citation Text

Zhe Liu, Hui Meng, Yongbin Zhang, Weiliang Duan, Ying Chen. Category Recognition of Three-Dimensional Fluorescence Spectra of Algae Based on LLE and Gold-SA-SVM[J]. Acta Optica Sinica, 2022, 42(4): 0430001

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites