高光谱与宽度学习检测草莓叶片含水率

doi:10.3964/j.issn.1000-0593(2025)06-1534-09

摘要
参考文献
相关文章 (15)

全文: PDF (21155 KB)
输出: BibTeX | EndNote (RIS)

摘要：水分是影响草莓生长发育的关键因素，对其栽培具有重要意义。传统的水分测量方法虽精确但繁琐且具有破坏性；高光谱成像技术（HSI）因其高效、非破坏性和多属性检测优势，成为植物水分检测的理想选择。HSI数据量大且信息冗余，深度学习方法虽然能提取数据深层特征，但对大规模标注数据的依赖性限制了其应用。为此，引入了宽度学习系统（BLS）以解决在小样本上的训练问题，提出了一种基于BLS的草莓叶片含水率检测方法。首先制备了健康及干旱胁迫的草莓叶片样本，获取其高光谱图像及水分含量数据。通过分析3种超参数调优方法和4种预处理算法，构建了BLS含水率测定模型，并与偏最小二乘回归（PLSR）、支持向量机回归（SVR）、梯度增强决策树回归（GBDTR）和残差神经网络（ResNet）等对比模型进行了性能评估。结果显示，BLS模型在测试集上的决定系数（R²）达到0.797 4，均方根误差（RMSE）为0.004 5，优于其他模型，比ResNet模型高0.039 4，证明其具有良好的泛化能力和预测准确性。此外，基于最优模型实现了草莓叶片含水率的可视化，通过生成含水率伪彩图直观展示草莓叶片的水分状态。研究结果表明，BLS模型在小样本的高光谱数据分析和草莓叶片含水率检测中具有可行性，为草莓叶片含水率的在线检测提供了科学依据。

关键词：草莓叶片；含水率；高光谱成像；宽度学习系统

Abstract：Moisture content is a critical factor influencing the growth and development of strawberries, and it holds significant importance for their cultivation. Traditional methods of moisture measurement, although precise, are cumbersome and destructive. Hyperspectral Imaging (HSI) has emerged as an ideal technique for plant moisture detection due to its efficiency, non-destructive nature, and multi-attribute detection capabilities. However, the large volume and redundancy of HSI data present challenges. While deep learning methods can extract deep features from the data, their reliance on large-scale annotated datasets limits their application. To address this issue, our study introduces a Broad Learning System (BLS) to solve the training problem with small sample sizes. It proposes a BLS-based method for detecting moisture content in strawberry leaves. The study first prepared samples of healthy and drought-stressed strawberry leaves, obtaining their hyperspectral images and moisture content data. By analyzing three hyperparameter tuning methods and four preprocessing algorithms, a BLS moisture determination model was constructed and its performance was evaluated against comparative models, including Partial Least Squares Regression (PLSR), Support Vector Machine Regression (SVR), Gradient Boosting Decision Tree Regression (GBDTR), and Residual Network (ResNet). The results showed that the BLS model achieved a coefficient of determination (R²_p) of 0.797 4 and a Root Mean Square Error (RMSE) of 0.004 5 on the test set, outperforming the other models and exceeding the ResNet model by 0.039 4, demonstrating its superior generalization ability and prediction accuracy. Additionally, the optimal model was used to visualize the moisture content in strawberry leaves by generating false-color maps, providing an intuitive display of the leaves' moisture status. The findings suggest that the BLS model is suitable for analyzing hyperspectral data and detecting moisture content in strawberry leaves using small samples, providing a theoretical basis for the online detection of moisture content in strawberry leaves.

Key words：Strawberry leaves; Moisture content; Hyperspectral imaging; Broad learning system

收稿日期: 2024-07-23 修订日期: 2025-02-06

中图分类号:

O433.4

基金资助: 山东省重大科技创新工程项目(2022CXGC020708)，浙江省重点研发计划项目(2025C02265)，国家自然科学基金项目（62105290）资助

通讯作者: 朱逢乐 E-mail: zhufl@zjut.edu.cn

作者简介: 李泽祺，2004年生，浙江工业大学机械工程学院本科生 e-mail: 478402481@qq.com

引用本文:

李泽祺，杨铮，周壮飞，彭继宇，朱逢乐，何青海. 高光谱与宽度学习检测草莓叶片含水率[J]. 光谱学与光谱分析, 2025, 45(06): 1534-1542.
LI Ze-qi, YANG Zheng, ZHOU Zhuang-fei, PENG Ji-yu, ZHU Feng-le, HE Qing-hai. Detection of Moisture Content in Strawberry Leaves Using Hyperspectral and Broad Learning System. SPECTROSCOPY AND SPECTRAL ANALYSIS, 2025, 45(06): 1534-1542.

链接本文:

https://www.gpxygpfx.com/CN/10.3964/j.issn.1000-0593(2025)06-1534-09 或 https://www.gpxygpfx.com/CN/Y2025/V45/I06/1534

[1] Sun H, Liu N, Wu L, et al. IFAC-PapersOnLine, 2018, 51(17): 443.
[2] ElMasry G, Wang N, ElSayed A, et al. Journal of Food Engineering, 2007, 81(1): 98.
[3] Li C C, Xiao Z, Liu Y H, et al. Agronomy, 2023, 13(1): 56.
[4] Asaari M S M, Mishra P, Mertens S, et al. ISPRS Journal of Photogrammetry and Remote Sensing, 2018, 138: 121.
[5] Li X, Chen Z, Wang J, et al. Sensors, 2023, 23(7): 3687.
[6] Aptoula E, Ozdemir M C, Yanikoglu B. IEEE Geoscience and Remote Sensing Letters, 2016, 13(12): 1970.
[7] Yang H, Zhang T, Yang W L, et al. International Agrophysics, 2023, 37(3): 225.
[8] An T, Yu H, Yang C S, et al. Journal of Food Process Engineering, 2020, 43(7): e13428.
[9] Lu Y Z, Chen D, Olaniyi E, et al. Computers and Electronics in Agriculture, 2022, 200: 107208.
[10] Ahmed S F, Alam M S B, Hassan M, et al. Artificial Intelligence Review, 2023, 56(11): 13521.
[11] Chen C L P, Liu Z. IEEE Transactions on Neural Networks and Learning Systems, 2018, 29(1): 10.
[12] Zhao G X, Wang X S, Kong Y, et al. Remote Sensing, 2021, 13(4): 583.
[13] Ma Y, Liu Z, Chen C L P. Applied Intelligence, 2022, 52(3): 2801.
[14] YE Rong-ke, KONG Qing-chen, LI Dao-liang, et al (叶荣珂，孔庆辰，李道亮, 等). Spectroscopy and Spectral Analysis (光谱学与光谱分析), 2022, 42(1): 164.
[15] Sinija V R, Mishra H N. Food and Bioprocess Technology, 2011, 4(1): 136.
[16] Li X L, Sun C J, Luo L B, et al. Computers and Electronics in Agriculture, 2015, 112: 28.
[17] Devos O, Downey G, Duponchel L. Food Chemistry, 2014, 148: 124.
[18] CHU Xiao-li, YUAN Hong-fu, LU Wan-zhen(褚小立，袁洪福，陆婉珍). Progress in Chemistry(化学进展), 2004, 16(4): 528.
[19] Yao K S, Sun J, Chen C, et al. Journal of Food Engineering, 2022, 325: 111024.
[20] Wu Y F, Peng S L, Xie Q, et al. Chemometrics and Intelligent Laboratory Systems, 2019, 185: 114.
[21] Amigo J M, Santos C. Preprocessing of Hyperspectral and Multispectral Images, in Data Handling in Science and Technology Elsevier, 2019, 37.
[22] Han J C, Zhang Z, Cao J, et al. Remote Sensing, 2020, 12(2): 236.
[23] Zhan D X, Mu Y Q, Duan W X, et al. Agriculture, 2023, 13(5): 1088.
[24] He K M, Zhang X Y, Ren S Q, et al. Deep Residual Learning for Image Recognition, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016: 770.
[25] Huang Q J, Mao J L, Liu Y. An Improved Grid Search Algorithm of SVR Parameters Optimization, 2012 IEEE 14th International Conference on Communication Technology, 2012: 1022.
[26] Bai Y, Shi L S, Zha Y Y, et al. Computers and Electronics in Agriculture, 2023, 215: 108349.
[27] Azimi S, Kaur T, Gandhi T K. Proceedings of 2020 25th International Conference on Pattern Recognition (ICPR), 2021, 8500.
[28] Zahedi S M, Moharrami F, Sarikhani S, et al. Scientific Reports, 2020, 10(1): 17672.
[29] Zahir S A D M, Jamlos M F, Omar A F, et al. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 2024, 304: 123273.
[30] Horler D N H, Dockray M, Barber J, et al. International Journal of Remote Sensing, 1983, 4(2): 273.
[31] Zhang Z J, Jiang D Y, Chang Q R, et al. Remote Sensing, 2023, 15(7): 1732.
[32] Yang L, Shami A. Neurocomputing, 2020, 415: 295.
[33] Haghbin N, Bakhshipour A, Zareiforoush H. Plant Methods, 2023, 19(1): 53.
[34] Mou Y, You X G, Xu D Q, et al. Chemometrics and Intelligent Laboratory Systems, 2014, 132: 168.
[35] Zhou J, Li W, Wang J X, et al. Physica A: Statistical Mechanics and its Applications, 2019, 534: 122370.
[36] Bejani M M, Ghatee M. Artificial Intelligence Review, 2021, 54(8): 6391.
[37] Jin J W, Liu Z L, Chen C L P. Science China Information Sciences, 2018, 61(11): 112209.
[38] Song Z Y, Yan S H, Zang Z Y, et al. IEEE Transactions on Terahertz Science and Technology, 2018, 8(5): 520.