Classification of Handkerchief Paper Brand Based on Spectral Data Fusion Technology

Ji Jiahua; Wang Jifen; He Xinlong

doi:10.3788/LOP202158.0330004

Journals >Laser & Optoelectronics Progress >Volume 58 >Issue 3 >Page 3300041 > Article

Laser & Optoelectronics Progress
Vol. 58, Issue 3, 3300041 (2021)

Classification of Handkerchief Paper Brand Based on Spectral Data Fusion Technology

Ji Jiahua, Wang Jifen^*, and He Xinlong

Author Affiliations

School of Investigation, People''s Public Security University of China, Beijing 100038, China

show less

DOI: 10.3788/LOP202158.0330004 Cite this Article Set citation alerts

Ji Jiahua, Wang Jifen, He Xinlong. Classification of Handkerchief Paper Brand Based on Spectral Data Fusion Technology[J]. Laser & Optoelectronics Progress, 2021, 58(3): 3300041 Copy Citation Text

show less

Fig. 1. Infrared spectra of handkerchief paper samples

Download full size

Fig. 2. Raman spectra of handkerchief paper samples

Download full size

Fig. 3. Trend chart of classification accuracy of 8 brands of handkerchief paper in different dimensions

Download full size

Fig. 4. Classification accuracy of 14 series of Xinxiangyin handkerchiefs in different dimensions

Download full size

Brand	Infrared spectrum			Raman spectrum
Brand	Accuracy/%	Recall/%	False positive/%	Accuracy/%	Recall/%	False positive/%
QF	92.1	66.7	5.3	96.8	66.7	0
SJ	85.7	66.7	11.1	98.4	88.9	0
WD	92.1	66.7	2.0	88.9	66.7	5.9
XBB	92.1	55.6	1.9	93.7	77.8	3.7
XS	90.5	83.3	8.8	93.7	83.3	5.3
XXY	92.1	77.8	5.6	90.5	77.8	7.4
XZL	93.7	66.7	3.5	98.4	83.3	0.0
YR	92.1	33.3	1.8	98.4	100	1.8

Table 1. LDA accuracy, recall rate and false positive rate of two spectra

Brand	Accuracy/%	Recall/%	False positive/%
QF	98.4	83.3	0
SJ	98.4	88.9	0
WD	92.1	75.0	3.9
XBB	95.2	88.9	3.7
XS	95.2	83.3	3.5
XXY	93.7	88.9	5.6
XZL	98.4	83.3	0
YR	96.8	83.3	1.8

Table 2. LDA accuracy, recall rate and false positive rate after spectral fusion

Component	Eigenvalue	Variance/%	Cumulative variance/%	Component	Eigenvalue	Variance/%	Cumulative variance/%
PC1	437.789	48.162	48.162	PC20	2.176	0.239	96.066
PC2	231.312	25.447	73.608	PC21	2.096	0.231	96.297
PC3	90.879	9.998	83.606	PC22	2.023	0.223	96.520
PC4	26.028	2.863	86.470	PC23	1.918	0.211	96.731
PC5	21.176	2.330	88.799	PC24	1.786	0.196	96.927
PC6	11.269	1.240	90.039	PC25	1.702	0.187	97.114
PC7	8.257	0.908	90.947	PC26	1.587	0.175	97.289
PC8	6.497	0.715	91.662	PC27	1.481	0.163	97.452
PC9	5.023	0.553	92.215	PC28	1.425	0.157	97.609
PC10	4.111	0.452	92.667	PC29	1.382	0.152	97.761
PC11	4.000	0.440	93.107	PC30	1.289	0.142	97.902
PC12	3.757	0.413	93.520	PC31	1.252	0.138	98.040
PC13	3.719	0.409	93.929	PC32	1.183	0.130	98.170
PC14	3.475	0.382	94.311	PC33	1.154	0.127	98.297
PC15	3.178	0.350	94.661	PC34	1.089	0.12	98.417
PC16	3.035	0.334	94.995	PC35	1.024	0.113	98.530
PC17	2.756	0.303	95.298	PC36	0.952	0.105	98.634
PC18	2.507	0.276	95.574	PC37	0.914	0.101	98.735
PC19	2.300	0.253	95.827

Table 3. Characteristic variance contribution rate of PCA

Dimension	Precision/%								Overall accuracy/%
Dimension	QF	SJ	WD	XS	XBB	XXY	XZL	YR	Overall accuracy/%
2	50.0	11.1	0	55.6	33.3	11.1	66.7	0	25.4
3	83.3	55.6	8.3	44.4	66.7	33.3	66.7	16.7	42.9
4	83.3	66.7	16.7	44.4	66.7	33.3	66.7	16.7	46.0
5	83.3	44.4	33.3	55.6	66.7	44.4	66.7	16.7	49.2
6	83.3	33.3	25.0	55.6	66.7	55.6	83.3	33.3	50.8
7	83.3	33.3	50.0	66.7	66.7	66.7	83.3	33.3	58.7
8	83.3	44.4	50.0	77.8	66.7	66.7	83.3	33.3	61.9
9	83.3	44.4	58.3	77.8	66.7	66.7	83.3	33.3	63.5
10	83.3	55.6	58.3	77.8	83.3	66.7	83.3	50.0	68.3
11	100	66.7	58.3	77.8	100	66.7	66.7	66.7	73.0
12	100	44.4	50.0	66.7	100	77.8	66.7	66.7	68.3
13	100	44.4	66.7	66.7	100	100	66.7	83.3	76.2
14	83.3	66.7	66.7	77.8	83.3	100	66.7	66.7	76.2
15	83.3	66.7	75.0	77.8	83.3	100	66.7	66.7	77.8
16	83.3	77.8	66.7	88.9	83.3	88.9	66.7	50.0	76.2
17	100	88.9	66.7	88.9	83.3	88.9	66.7	83.3	82.5
18	100	100	83.3	88.9	100	100	100	83.3	93.7
19	100	100	91.7	88.9	100	88.9	83.3	100	93.7
20	100	100	83.3	77.8	100	88.9	100	100	92.1
21	100	100	83.3	88.9	100	88.9	83.3	100	92.1
22	100	100	83.3	88.9	100	100	100	100	95.2
23	100	88.9	83.3	88.9	100	100	100	100	93.7
24	100	100	75.0	88.9	100	100	100	100	93.7
25	100	100	83.3	88.9	100	100	100	100	95.2
26	100	100	91.7	88.9	100	100	100	100	96.8
27	100	100	91.7	88.9	100	100	100	100	96.8
28	100	100	91.7	88.9	100	100	100	100	96.8
29	100	100	91.7	88.9	100	100	100	100	96.8
30	100	100	91.7	88.9	100	100	100	100	96.8
31	100	100	100	88.9	100	100	100	100	98.4
32	100	100	100	88.9	100	100	100	100	98.4
33	100	100	100	88.9	100	100	100	100	98.4
34	100	100	100	100	100	100	100	100	100

Table 4. Accuracy of LDA in different dimensions

Material	QF	SJ	WD	XS	XBB	XXY	XZL	YR
JC1						√
JC2			√

Table 5. Comparison results of material and sample

Ji Jiahua, Wang Jifen, He Xinlong. Classification of Handkerchief Paper Brand Based on Spectral Data Fusion Technology[J]. Laser & Optoelectronics Progress, 2021, 58(3): 3300041

Download Citation

Set citation alerts for the article

Tools

Set citation alerts for the article

Save the article for my favorites

Paper Information