基于透射光谱的玉米叶片含水率快速检测仪研究

为快速、无损检测玉米叶片含水率,根据透射光谱原理设计一款便携式植物叶片水分快速检测仪。检测仪主要由数据采集节点和数据接收节点两部分组成,数据采集节点采用夹持叶室结构,由信号采集、处理和发送模块组成,数据接收节点由信号接收模块和PDA组成。测量时玉米叶片被放入采集节点的夹持叶室,两路LED主动光源(890 nm和980 nm)发光照射叶片,采用PIN型Si光电传感器在叶片的另一面进行透射光探测,透射光信号经调理电路放大、滤波后,通过ZigBee网络发送至数据接收节点。根据采集的光信号计算透射率(T890和T980)、比值植被指数(RVI)和调整型归一化差异水分指数(MNDWI)等参数,分析了各植被指数与不同叶位含水率之间的相关性,结果显示仪器应用的最佳叶位为完全展开叶片的倒二叶中部,含水率检测范围为70%~80%,且分辨率为0.3%。选取了T890、T980和MNDWI建立了含水率检测模型,其R_C~2为0.854,R_V~2为0.849,RMSE为0.010 3。

Rapid Detection of Moisture Content in Maize Leaves Based on Transmission Spectrum

In order to detect the moisture content of maize leaves quickly and nondestructively, a portable plant leaf moisture detector was developed based on near infrared spectroscopy and wireless sensor network. The structure was designed according to the principle of transmission spectrum. The device could be divided into data acquisition node and data receiving node. And the ZigBee network was used to transfer the data between them. There were three modules in the data acquisition node, including a signal acquisition module, a signal processing module and a signal transmission module. The data acquisition node mainly completed the acquisition, amplification and filtering of the spectral data, and then sent it to the PDA through the master chip JN5168. The data receiving node was designed with a data receiving module and a PDA. The signal acquisition module was the key component with a light source and a sensor. The active LED lights in near-infrared band (890nm and 980nm) sent light, and the PIN type photodiode was used to measure the transmission signal. After the signal amplification and filtering, it was sent to the data receiving node by the ZigBee network. The parameters were calculated, including T890, T980, ratio vegetation index (RVI), modified normalized difference water vegetation index (MNDWI) and so on. The field test was conducted. The correlation between the parameters and water content was analyzed. The best application location of maize leaves was recommended. The results showed that the best detecting range was 70%~80% and the resolution was 0.3%. The detection model was established by T890, T980 and MNDWI, with R2C=0.854, R2V=0.849, and root mean square error of 0.0103. It could help to detect leaf moisture content nondestructively in the field.