Spectroscopy and Spectral Analysis, Volume. 42, Issue 4, 1250(2022)
Classification Method of Coal and Gangue Based on Hyperspectral Imaging Technology
Figures & Tables(8)
Fig. 2. Hyperspectral imaging system (a) and process of spectrum extraction (b)
Fig. 3. The original spectral curves in the range of 474~940 nm (a) and 1 235~2 477 nm (b)
Fig. 4. The process of variable selection by CARS in the spectral ranges of (a) Vis/NIR and (b) NIR
Fig. 5. The results of variable selection by SPA over the spectral ranges of (a), (c) Vis/NIR and (b), (d) NIR
Fig. 6. Grayscale images of some samples and corresponding classification visualization
red: Coal; blue: Gangue
Table 1. Comparison of different classification models based on the full-band spectra
View tableView in ArticleTable 1. Comparison of different classification models based on the full-band spectra
Spectral range Number of variables Model sensitivity specificity accuracy SVM 1.000 0 0.956 5 0.979 2 Vis/NIR 600 KNN 1.000 0 0.956 5 0.979 2 PLS-DA 1.000 0 0.956 5 0.979 2 SVM 1.000 0 0.960 9 0.981 3 NIR 196 KNN 0.956 0 0.956 5 0.956 3 PLS-DA 1.000 0 0.987 0 0.993 8
Table 2. The prediction results of different classification models based on characteristic wavelengths
View tableView in ArticleTable 2. The prediction results of different classification models based on characteristic wavelengths
Spectral
rangeVariable selection
methodsNumber of
variablesModel sensitivity specificity accuracy Vis/NIR SVM 1.000 0 0.930 4 0.966 7 CARS 5 KNN 0.976 0 0.930 4 0.954 2 PLS-DA 0.956 0 0.830 4 0.895 8 SVM 1.000 0 0.965 2 0.983 3 SPA 3 KNN 1.000 0 0.956 5 0.979 2 PLS-DA 0.776 0 0.869 6 0.820 8 Vis/NIR SVM 0.888 0 0.995 7 0.939 6 CARS 3 KNN 0.988 0 0.991 3 0.989 6 PLS-DA 1.000 0 0.965 2 0.983 3 SVM 0.960 0 0.947 8 0.954 2 SPA 3 KNN 0.960 0 0.943 5 0.952 1 PLS-DA 0.932 0 0.917 4 0.925 0
Tools
Get Citation
Copy Citation Text
Lian-jie LI, Shu-xiang FAN, Xue-wen WANG, Rui LI, Xiao WEN, Lu-yao WANG, Bo LI. Classification Method of Coal and Gangue Based on Hyperspectral Imaging Technology[J]. Spectroscopy and Spectral Analysis, 2022, 42(4): 1250
Paper Information
Category: Research Articles
Received: Mar. 14, 2021
Accepted: --
Published Online: Jul. 25, 2023
The Author Email: LI Lian-jie (lilianjie0063@link.tyut.edu.cn)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20