玉米叶绿素含量高光谱反演的线性模型研究

Research on Linear Model of High Spectral Retrieval of Chlorophyll Content in Corn

叶绿素含量是衡量植被生长状况的一个重要指标,高光谱数据具有较高的光谱分辨率,利用其光谱信息建立叶绿素含量的关系模型,已成为监测植被长势的一种有效手段。传统叶绿素含量线性回归模型的输入因子是植被特征提取参数,由于高光谱数据波段间的冗余度较高,导致一般的线性模型的反演精度较低。主成分分析可以减少数据的维数,简化网络结构,得出能反映原始信息的综合变量。本文以盆栽玉米为研究对象,利用植被特征和主成分分析方法提取光谱反演参数,根据所提取的参数建立玉米叶片叶绿素含量的一元线性和多元线性回归模型。结果表明,利用绿峰峰值和近红外反射率均值两参数可在一元线性模型中较好地反演玉米叶片叶绿素含量;而利用分波段提取的主成分能够在多元线性回归模型中更好地反演叶绿素含量,反演精度较高。

The chlorophyll content is an important indicator to measure the vegetation growth status, and the hyperspectral data have fine spectral resolution, so to establish the chlorophyll content model by using the spectral information has become an effective means for monitoring vegetation growth. The input factors of the conventional linear regression model for chlorophyll content were the feature extraction parameters of vegeta- tion, so high redundancy between the bands of hyperspeetral data resulted in lower accuracy of the general lin- ear model. The principal component analysis （PCA） can effectively reduce the data dimensions and simplify the network structure to obtain comprehensive variables reflecting the original information. Using the potted corn as research object, the linear and multiple linear regression models of corn leaf chlorophyll content were established with the spectral parameters extracted by vegetation characteristics and principal component method. The results showed that the green peak value and the mean of near - infrared refleetivity were better for the inversion of chlorophyll content in corn leaves with linear mode, while the principal components extracted from different brands were better for multiple linear regression model.