基于反射光谱的PCA及BP神经网络法预测甘蔗叶片叶绿素含量

叶绿素是植物进行光合作用的重要色素，叶绿素含量可以作为评价植物生长状况的重要参数。本研究基于甘蔗叶片的反射光谱，利用PCA及BP神经网络算法，建立了甘蔗叶片的叶绿素含量预测模型。PCA算法可以在尽可能少地丢失有用光谱信息的前提下，降低输入光谱矩阵的维数，最大限度地减少冗余信息。BP神经网络算法因其良好的非线性逼近能力可大大提高该模型的预测精度。研究发现：基于PCA和BP算法建立的叶绿素含量预测模型，其预测值与实测值之间的R2达0.8929，表明该模型具有较高的预测能力。

Chlorophyll Content Prediction for Sugarcane Leaves Based on Spectral Reflectance with PCA and BP Neural Network Algorithm

Chlorophyll is the primary pigment for plants photosynthesis, so it iscommonly considered as an important parameter to evaluate the growth status of plants. Based on the spectra reflectance, a chlorophyll prediction model for sugarcane leaves with PCA and BP neural network algorithm was built in this study. PCA algorithm couldminimize the redundant information; reduce the dimension of the input spectra matrix without losing too much useful spectral information. BP neural network algorithm could also improve the accuracy of the prediction model because of its nonlinear approximation capability. The research found that the chlorophyll prediction model builtwith PCA and BP algorithm had R2 of 0.8929 between its prediction values and measured values,indicating a high predictive ability.