基于偏最小二乘法的樟树叶片叶绿素含量高光谱估算

Hyperspectral Estimation of Camphor Tree Leaf Chlorophyll Content Based on Partial Least Squares Regression

在BRDF测试系统环境下利用ASD便携式野外光谱仪采集樟树叶片光谱，并用UV2450－紫外可见分光光度计对观测叶片进行叶绿素含量测定．考虑到植物色素（叶绿素和类胡萝卜素）对叶片反射光谱的影响主要体现在可见光波段，选取400～900nm范围波段光谱反射率与叶片叶绿素含量反演偏最小二乘法（PLS）模型，其中29个样本用于建模，10个样本用于验证，结果表明：当主成分个数为4时，PLS模型具有最佳的效果，4个主成分累计解释了99．9l％的自变量信息和89．71％的因变量信息，此外，PLS模型能够充分利用高光谱信息，具有较高的精度和稳定性．通过与原始光谱和一阶导数光谱拟合的估测模型进行对比分析，得出PLS模型无论是从建模样本精度还是验证的误差方面均优于这两种传统的模型，适合于利用高光谱数据进行叶绿素含量的估测．

Under BRDF ASD test system conditions, it acquired the spectrum of Camphor tree leaves using a portable field spectrometer and determined the chlorophyll content in the studied leav- es using a UV2450 spectrophotometer. Considering the plant pigments （chlorophyll and carotenoids） affect leaf spectral reflectance in visible light band chiefly, so this paper selected 400 - 900 nm band and leaf chlorophyll content to inversion model of partial least squares. Then 29 samples were use to model, and other 10 samples were used for model verification. The results show that PLS model is suited when the number of principal components is 4, These three principal compo- nents explain 99.91% of independent variables information and 89.71% of dependent variables in- formation. PLS model can make full use of the information of hyperspectral data, and hence have higher accuracy and stability. Compare to the original spectrum and first derivative spectral estimation model, the results show that PLS model is better than the two traditional models no matter on the accuracy of modeling samples or the error of validation samples, indicating that it is suitable for the estimation of chlorophyll content by using hyperspectral data.