Spectroscopy and Spectral Analysis, Volume. 40, Issue 12, 3705(2020)

Research Progress of Spectroscopy in the Detection of Soil Moisture Content

Xin-xing LI1,1、*, Bu-wen LIANG1,1, Xue-bing BAI1,1, and Na LI1,1
Author Affiliations
  • 1[in Chinese]
  • 11. Beijing Laboratory of Food Quality and Safety, College of Information and Electrical Engineering, China Agricultural University, Beijing 100083, China
  • show less
    Figures & Tables(3)
    • Table 1. Characteristic analysis of spectral pretreatment technology

      View table
      View in Article

      Table 1. Characteristic analysis of spectral pretreatment technology

      方法原理特点参考文献
      Savitzky-Golay平滑算法
      Savitzky-Golay smoothing algorithm(SG)
      Xi*=j=-rlXi+jWjj=-llWj
      Wj是移动窗口平滑中的权重因子(窗口长度2l+1)
      不受样本数据限制, 适用于光谱信号的稳定去噪, 对高频噪音消除效果较好[23]
      多元散射校正
      Multiplicative scatter correction(MSC)
      Xi*=X¯i-bimi
      mibi表示Xi与线性回归后的相对偏移系数和平移量
      消除分布不均匀和颗粒大小造成的散射影响[24]
      微分处理
      Differential algorithm
      Xi*=Xi+j-Xig
      G为差分宽度
      消除背景干扰的影响, 突出光谱变化规律[25-26]
    • Table 2. Comparative analysis of feature extraction algorithms

      View table
      View in Article

      Table 2. Comparative analysis of feature extraction algorithms

      方法特点适用范围参考文献
      包络线消除法
      (continuum-removal, C-R)
      具有降低非目标光谱数据影响, 扩大水分吸收特征信息的优势适用于单一敏感波段特征提取
      主成分分析
      (principal component analysis, PCA)
      将众多且较强相关性的光谱数据转化为维数较少的特征信息适用于多因素的光谱特征提取, 使其主要的特征信息保留[25]
      遗传算法
      (genetic algorithm, GA)
      具有全局搜索手段和的可拓展性多与机器学习结合提取信息
    • Table 3. Characteristic analysis of spectral modeling methods

      View table
      View in Article

      Table 3. Characteristic analysis of spectral modeling methods

      类型建模方法土壤类型模型精度测试集参考文献
      一元线性一元线性y=1.271x-0.015 1浙江省红壤验证集r=0.966 5, RMSE=0.012 1
      一元线性y=17.706-2.167x北京郊区褐土建模集R2=0.935, 验证集R2=0.936
      多元线性多元线性模型
      (multiple regression analysis, MRA)
      新疆干旱区土壤多种线性模型, 建模集
      最高精度可达R2=0.963
      偏最小二乘
      (partial least-square method, PLS)
      银川地区半干旱土壤R2=0.976
      非线性支持向量机
      (support vector machine, SVM)
      新疆绿洲干旱区土壤建模集R2=0.86, RMSE=4.16
      验证集R2=0.92, RMSE=2.83
      [30]
      高斯核函数支持向量机德国半湿润性黏土建模集R2=0.983, RMSEP=0.457[25]
      多关联向量机
      (relevance vector machines, RVM)
      犹他州半干旱土壤Bubble验证集RMSE=0.06
      Laplace验证集RMSE=0.08
      Cauchy验证集RMSE=0.10
      [31]
      卷积神经网络
      (convolutional neural networks, CNN)
      江苏句容市湿润黏土建模集R2=0.968, RMSE=0.57
      验证R2=0.956, RMSE=0.804
    Tools

    Get Citation

    Copy Citation Text

    Xin-xing LI, Bu-wen LIANG, Xue-bing BAI, Na LI. Research Progress of Spectroscopy in the Detection of Soil Moisture Content[J]. Spectroscopy and Spectral Analysis, 2020, 40(12): 3705

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: Research Articles

    Received: Oct. 15, 2019

    Accepted: --

    Published Online: Jun. 18, 2021

    The Author Email: LI Xin-xing (lxxcau@cau.edu.cn)

    DOI:10.3964/j.issn.1000-0593(2020)12-3705-06

    Topics