基于可见-近红外光谱技术的油菜叶片叶绿素含量无损检测研究

    为了快速无损获取油菜叶片叶绿素含量信息,试验研究了油菜叶片的可见-近红外反射光谱特性与叶绿素含量之间的定量关系.试验采集140个油菜叶片样本,其中70个样本用于建模,另外70个样本用于模型预测.光谱曲线扫描采用美国USB4000光纤光谱仪,叶绿素含量值采用日本Minolta 公司生产的SPAD-502仪测定.实验发现,波段范围680～730 nm处的光谱吸光度与油菜叶片叶绿素含量之间具有显著相关性.同时发现油菜叶片厚度对建模预测精度有较大影响.试验首先用待定系数法构造叶绿素含量预测方程;然后用标准遗传算法对其进行参数优化.试验确定最优光谱范围是696.82～716.53 nm.不考虑叶片厚度时,建模和预测关联度r分别是0.4823 和0.5649.考虑叶片厚度校正后,建模和预测关联度r分别提高到0.8936 和0.9178.说明基于可见-近红外反射光谱技术实现油菜叶片叶绿素含量快速无损检测是可行的.

Nondestructive detection of rape leaf chlorophyll level based on Vis/NIR spectroscopy

In order to measure chlorophyll level of rape leaf nondestructively and instantly, the relation between visual and near infrared(Vis/NIR) reflectance spectra of leaves and their SPAD values was examined. In the test, 140 rape leaf samples were selected. Among them, 70 samples were used for model calibration and other 70 were for model verification. Each leaf sample was spectrally scanned by modular spectrometer USB4000, Ocean Optics, USA. Chlorophyll level of each rape leaf was measured by SPAD-502 meter, Minolta Camera, Japan. By the observation of spectral curves, the spectral range between 680 nm and 730 nm was found significant for predictive modeling of chlorophyll level of rape leaf. It was also found necessary to put leaf thickness into consideration. The procedure of shaping the predictive model was as follows: firstly, prediction equation of rape leaf chlorophyll level was created with uncertain parameters;secondly, a standard genetic algorithm (GA) was designed for parameter optimization. As the result of the GA calculation, the optimal spectral range was narrowed within 696.82 nm and 716.53 nm. Compared with the r=0.4823 for calibration set and r=0.5649 for verification set without concerns of leaf thickness, the effect of leaf thickness on the spectral modeling was significant: the r of calibration set and verification set was improved as high as 0.8936 and 0.9178, respectively. The test shows that it is practical to use Vis/NIR reflection spectrometer for the nondestructive determination of chlorophyll level of rape leaf.