基于机器学习的柑橘叶绿素含量相关性分析研究

Correlation Analysis of Citrus Chlorophyll Content based on Machine Learning

以广西壮族自治区桂林市雁山区柑橘种植区为研究区,以无人机高光谱影像为数据源,采用随机森林、支持向量机、Xgboost三种机器学习算法分别对叶绿素含量进行相关性分析。结果表明,六种机器学习模型通过引入多个与SPAD值显著相关的光谱特征参数,比单个光谱特征参数构建的估算模型效果要好;同时Xgboost回归接近均好于RF模型,说明逐步回归、SVM的验证模型的稳定性比RF高,模型的预测精度决定系数达到0.930。

Liutang Town,Yanshan District,Guilin City,Guangxi Zhuang Autonomous Region was used as the research area,and UAV hyperspectral images were used as the data source,and three machine learning algorithms were used to analyze the chlorophyll content by random forest,support vector machine and Xgboost.The results show that the six machine learning models have better effect than the estimation models constructed by introducing multiple spectral feature parameters that are significantly correlated with SPAD values.At the same time,the Xgboost regression approach was better than that of the RF model,indicating that the stability of the stepwise regression and SVM verification model was higher than that of RF,and the prediction accuracy coefficient of the model reached 0.930.