对灰霉病胁迫下番茄叶片中叶绿素含量(SPAD)的高光谱图像信息进行了研究。 首先获取380～1 030 nm波段范围内健康和染病番茄叶片的高光谱图像, 然后基于ENVI软件处理平台提取高光谱图像中感兴趣区域的光谱信息, 经平滑(Smoothing)、 标准化(Normalize)等预处理后, 建立了基于Normalize预处理的偏最小二乘回归(PLSR)和主成分回归(PCR)模型。 再基于PLSR获得的4个变量建立反向传播神经网络(BPNN)和最小二乘-支持向量机(LS-SVM)模型。 4个模型中, LS-SVM的预测效果最好, 其决定系数R2为0.901 8, 预测集均方根误差RMSEP为2.599 2。 结果表明, 基于健康和染病番茄叶片的高光谱图像响应特性检测叶绿素含量(SPAD)是可行的。

Hyperspectral imaging feature of chlorophyll content (SPAD) in tomato leaves stressed by grey mold was studied in the present paper. Hyperspectral imagings of healthy and infected tomato leaves were obtained by hyperspectral imaging system from 380 to 1 030 nm and diffuse spectral response of region of interest (ROI) from hyperspectral imaging was extracted by ENVI software, then different preprocessing methods were used including smoothing and normalization etc. The partial least squares regress (PLSR) and principal component regress (PCR) models were developed for the prediction of SPAD value in tomato leaves based on normalization preprocessing method, then the back-propagation neural network (BPNN) and least squares-support vector machine(LS-SVM)models were built based on the four variables suggested by PLSR model. Among the four models, LS-SVM model was the best to predict SPAD value and the coefficient of determination (R2) was 0.901 8 with the root mean square error of prediction (RMSEP) of 2.599 2. It was demonstrated that chlorophyll content (SPAD) in healthy and infected tomato leaves can be effectively detected by the hyperspectral imaging technique.