Classification of wheat grain varieties using terahertz spectroscopy and convolutional neural network

作者全名：Chen, Fang; Shen, Yin; Li, Guanglin; Ai, Ming; Wang, Liang; Ma, Huizhen; He, Wende

作者地址：[Chen, Fang] Luzhou Vocat & Tech Coll, Luzhou 646000, Peoples R China; [Shen, Yin] Chongqing Med Univ, Coll Med Informat, Chongqing 400016, Peoples R China; [Li, Guanglin] Southwest Univ, Coll Engn & Technol, Chongqing 400715, Peoples R China; [Ai, Ming] Chongqing Med Univ, Affiliated Hosp 1, Dept Psychiat, Chongqing 400016, Peoples R China; [Wang, Liang] Chongqing Med Univ, Affiliated Hosp 1, Dept Neurol, Chongqing 400016, Peoples R China; [Ma, Huizhen] ChongQing Acad Anim Sci, Prataculture Res Inst, Chongqing 402460, Peoples R China; [He, Wende] Sichuan Vocat Coll Chem Ind, Luzhou 646000, Peoples R China

通信作者：Shen, Y (通讯作者)，Chongqing Med Univ, Coll Med Informat, Chongqing 400016, Peoples R China.; Li, GL (通讯作者)，Southwest Univ, Coll Engn & Technol, Chongqing 400715, Peoples R China.

来源：JOURNAL OF FOOD COMPOSITION AND ANALYSIS

ESI学科分类：AGRICULTURAL SCIENCES

WOS号：WOS:001202923100001

JCR分区：Q2

影响因子：4

年份：2024

卷号：129

期号：　

开始页：　

结束页：　

文献类型：Article

关键词：Terahertz spectral; Convolutional neural network; Wheat varieties; Qualitative evaluation

摘要：Wheat quality and quantity differ in diverse climates. Therefore, it is essential to identify the variety before purchasing and warehousing. In this study, a study on variety discrimination for 12 wheat varieties (stronggluten wheat, medium -gluten wheat, weak -gluten wheat) using the Terahertz time -domain spectroscopy (THzTDS) technology in combination with a Convolutional neural network (CNN). Firstly, the original Time -domain spectra (TDS) of wheat in the range of 0.1-2.0 THz were acquired, and the Frequency domain spectra (FDS), the absorption coefficient spectra in the range of 0.2-1.0 THz were obtained through Fourier Transform. Then, Competitive adaptive reweighted sampling (CARS) algorithms were applied to screen the feature spectrum. Finally, the Support vector machine (SVM), the Least square support vector machine (LS-SVM), the Backpropagation neural networks (BPNN) and CNN models were constructed using feature spectral data. By comparing the four models, it was found that the calibration set accuracy and prediction set accuracy of the CNN model reached 98.7% and 97.8% respectively, with an error recognition rate of only 2.2%. The research results show that combining THz-TDS technology with CNN has the advantages of accurate recognition and high efficiency. It can identify different wheat varieties and can be used for seed classification and quality detection.

基金机构：Intelligent Medicine Research Project of Chongqing Medical University [ZHYXQNRC202207]; Program for Youth Innovation in Future Medicine, Chongqing Medical University [W0138]; Future Medical Youth Innovation Team Development Support Plan of Chongqing Medical University [W0099]; Chongqing Natural Science Foundation of China [CSTB2022NSCQ-MSX0886]; Natural Science Foundation in Jiangxi Province of China [20202BAB212007]

基金资助正文：We would also like to thank Beijing Hybrid Wheat Engineering Research Center, Beijing Academy of Agriculture and Forestry Sciences for providing the Sichuan Pepper used in this research. This work was supported by the Intelligent Medicine Research Project of Chongqing Medical University (ZHYXQNRC202207) , the Program for Youth Innovation in Future Medicine, Chongqing Medical University (No. W0138) , the Future Medical Youth Innovation Team Development Support Plan of Chongqing Medical University (No. W0099) , Chongqing Natural Science Foundation of China (No. CSTB2022NSCQ-MSX0886) and Natural Science Foundation in Jiangxi Province of China (20202BAB212007) .