橡胶树叶片钾素含量高光谱估算模型研究  

Study on Hyperspectral Estimation Models for Potassium Content of Rubber Tree Leaves

作　　者：黎小清[1] 陈桂良[1] 许木果 丁华平[1] 刘忠妹 杨春霞[1] LI Xiao-qing;CHEN Gui-liang;XU Mu-guo;DING Hua-ping;LIU Zhong-mei;YANG Chun-xia(Yunnan Institute of Tropical Crops,Yunnan Jinghong 666100,China)

机构地区：[1]云南省热带作物科学研究所,云南景洪666100

出　　处：《西南农业学报》2020年第4期769-774,共6页Southwest China Journal of Agricultural Sciences

基　　金：云南省科研院所技术开发研究专项(2018DC031);云南省热带作物科技创新体系建设资金(RF2019-5,RF2019-6);云南省万人计划“产业技术领军人才”专项(云发改人事[2014]1782号)。

摘　　要：【目的】文章尝试建立橡胶树叶片钾素含量高光谱估算模型。【方法】基于涵盖2个品种和6个割龄的橡胶树叶片钾素含量和光谱反射率,分别采用偏最小二乘法(PLS)和偏最小二乘-支持向量机(PLS-SVM)构建了橡胶树叶片钾素含量高光谱估算模型。采用多元散射校正(MSC)、标准正态变换(SNV)、SG平滑(SG)、一阶导数(FD)和二阶导数(SD)等预处理方法或方法组合对叶片原始光谱反射率进行预处理,通过筛选得到最佳预处理模式为SNV+SG+FD的组合方法,其中,SG平滑点数为7,多项式次数为1或2。【结果】利用验证集对PLS模型和PLS-SVM模型进行验证,结果显示,决定系数R^2分别为0.843和0.868,均方根误差RMSE分别为1.331和1.162 g/kg,相对分析误差RPD分别为2.414和2.764。【结论】2种模型均具有良好的估算效果,PLS-SVM模型具有更高的估算精度。【Objective】This paper tried to establish a hyperspectral estimation model for potassium content in rubber leaves.【Method】Based on potassium content and spectral reflectance of rubber tree leaves,two varieties and six tapping ages,the hyperspectral estimation model of potassium of rubber tree leaves was established using partial least squares(PLS)and partial least squares-support vector machine(PLS-SVM),respectively.Multiplicative scatter correction(MSC),standard normal variate(SNV),Savitzky-Golay smoothing(SG),first derivative(FD),second derivative(SD)and the combination of which were used to preprocess the spectral reflectance of leaves.Optimal Preprocessing model was the combination of SNV,SG and FD,and the SG smooth points number was 7,the polynomial degree was first or two.【Result】The PLS model and PLS-SVM model were validated by validation set,and the results showed that the R^2were 0.843 and 0.868,the RMSE were 1.331 and 1.162 g/kg,and the RPD were 2.414 and 2.764,respectively.【Conclusion】Both models had good estimation effect.The PLS-SVM model had the higher estimation accuracy.

关 键 词：高光谱 橡胶树叶片 钾素 估算模型 偏最小二乘法 支持向量机 

分 类 号：O657.3[理学—分析化学] S129[理学—化学]