Spectroscopy and Spectral Analysis, Volume. 42, Issue 3, 873(2022)
Mineral Spectra Classification Based on One-Dimensional Dilated Convolutional Neural Network
Figures & Tables(7)
Fig. 1. Part of the mineral spectra
(a): Muscovite; (b): Dolomite; (c): Calcite; (d): Kaolinite
Fig. 2. Schematic of one-dimensional dilated convolution
(a): Standard convolution; (b): Dilated convolution
Fig. 3. The structure of one-dimensional dilated convolutional neural network
Fig. 4. Model performances with different convolution kernel types
Table 1. The parameters of 1D-DCNN
View tableView in ArticleTable 1. The parameters of 1D-DCNN
层序号 层类别 卷积核大小 卷积核数量 步长 1 Input layer 511×1 2 DC1 5×1 64 1 3 S1 3×1 2 4 DC2 3×1 64 1 5 S2 3×1 2 6 DC3 3×1 64 1 7 FC1 256 outputs 8 FC2 64 outputs 9 Output layer 4 outputs
Table 2. 1D-DCNN model discrimination results of different number of iteration
View tableView in ArticleTable 2. 1D-DCNN model discrimination results of different number of iteration
迭代次数 准确率/% 10 54.27 50 92.16 100 96.98 150 98.32 200 99.40 250 98.99
Table 3. The classification accuracies of different algorithms
View tableView in ArticleTable 3. The classification accuracies of different algorithms
分类算法 准确率/% SVM 97.94 BP 98.65 1D-DCNN 99.32
Tools
Get Citation
Copy Citation Text
Qing-lin TIAN, Bang-jie GUO, Fa-wang YE, Yao LI, Peng-fei LIU, Xue-jiao CHEN. Mineral Spectra Classification Based on One-Dimensional Dilated Convolutional Neural Network[J]. Spectroscopy and Spectral Analysis, 2022, 42(3): 873
Paper Information
Category: Orginal Article
Received: Feb. 26, 2021
Accepted: Jun. 15, 2021
Published Online: Apr. 19, 2022
The Author Email: TIAN Qing-lin (736924158@qq.com)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20