Hyperspectral Estimation of Selenium Content in Selenium-Rich Tea Based on Feature Selection and Machine Learning
WEN Zhu1, GUO Song1, SHU Tian1, ZHAO Long-cai2, 3
1. Guizhou Agricultural Science and Technology Information Institute, Guiyang 550006, China
2. College of Natural Resources and Environment, Key Laboratory of Plant Nutrition and Agro-Environment in Northwest China, Ministry of Agriculture and Rural Affairs, Yangling 712100, China
3. College of Natural Resources and Environment, Northwest A&F University, Yangling 712100, China
Abstract：Selenium is one of the important nutrient indices in selenium-rich tea, and its content determines the economic and nutritional value of selenium-rich tea. Hyperspectral remote sensing inversion technology has the characteristics of non-destructive, real-time, and rapid monitoring. This study utilizes the selenium content in selenium-rich tea from the Nangong River tea garden in Kaiyang County, Guizhou Province, and corresponding canopy non-imaging hyperspectral data as source data. The Savitzky-Golay second-order smoothing filter was used to preprocess the primary spectrum, and the potential of the primary spectral data was explored through first-order derivative transformation and continuum removal transformation. The independent variables for the modeling were obtained using a band elimination combination and various feature selection algorithms. Multiple inversion models of selenium content in tea were constructed using different algorithms. The results showed that: (1) the combination of spectral transformation and spectral index could enhance the ability of retrieving selenium content from the primary spectrum. (2) SPA was better than UVE overall; Continuum removal spectrum was superior to the primary spectrum and the first derivative spectrum. (3) The accuracy of the multi-factor model was better than that of the factor model, and the performance of ELMR in the multi-factor model was the best. Among all the models, the SPA-ELMR model under the continuum removal spectrum had the highest accuracy. The coefficient of determination (R2) and normalized root mean square error (nRMSE) of this model were 0.689 and 18.869%, respectively, and the corresponding verification R2 and nRMSE were 0.627 and 20.429%, respectively. In this study, the response relationship between selenium content in tea and spectral reflectance at specific growth stages was discussed. A single-factor inversion model and a multi-factor inversion model with appropriate accuracy were constructed, providing a theoretical basis for the rapid and non-destructive monitoring of selenium content in tea. Also, they provided some technical support for the digital construction of tea gardens.
[1] Zhang T, Qi M, Wu Q, et al. Frontiers in Nutrition, 2023, 10: 1183487.
[2] Wang K, Fang Q, He P, et al. Trends in Food Science & Technology, 2024, 145: 104356.
[3] Kolackova T, Sumczynski D, Bednarik V, et al. Journal of Food Composition and Analysis, 2021, 97: 103792.
[4] Zhang Y, Xiao J, Yan K, et al. Agronomy, 2023, 13(8): 2163.
[5] Tu Y, Bian M, Wan Y, et al. PeerJ, 2018, 6: e4858.
[6] Sonobe R, Hirono Y, Oi A. Plants, 2020, 9(3): 368.
[7] YANG Bao-hua, GAO Yuan, WANG Meng-xuan, et al(杨宝华，高 远，王梦玄，等). Spectroscopy and Spectral Analysis(光谱学与光谱分析), 2021, 41(3): 936.
[8] Kang Y S, Ryu C, Suguri M, et al. Food Chemistry, 2022, 370：130987.
[9] He Z, Wu K, Wang F, et al. Remote Sensing, 2023, 15(4)：1100.
[10] Ren G, Yin L, Wu R, et al. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 2024, 308: 123740.
[11] Jiang J, Ji H, Yan Y, et al. Computers and Electronics in Agriculture, 2024, 225：109358.
[12] Tran N K, Kuehle L C, Klau G W. Pattern Recognition Letters, 2024, 178: 69.
[13] Wang J, Lu S, Wang S H, et al. Multimedia Tools and Applications, 2022, 81(29): 41611.
[14] Awadallah M A, Al-Betar M A, Doush I A, et al. Archives of Computational Methods in Engineering, 2023, 30(5): 2831.
[15] ZHAO Long-cai, LI Fen-ling, CHANG Qing-rui(赵龙才，李粉玲，常庆瑞). Transactions of the Chinese Society for Agricultural Machinery(农业机械学报), 2023, 54(2): 1.
[16] Tian C, Lu Y, Xie H, et al. Scientific Reports, 2025, 15(1): 3895.
[17] Chen X, Li F, Shi B, et al. Agronomy, 2023, 13(3): 783.
[18] Ta N, Chang Q, Zhang Y. Remote Sensing, 2021, 13(19): 3902.
