基于多光谱成像技术的玉米氮素营养诊断方法研究

Research on Nutrition Diagnosis Method of Maize Nitrogen Based on Multispectral Imaging Technique

为快速、无损地获取寒地玉米作物养分信息,利用多光谱成像技术开展了大田玉米氮素营养诊断研究。采用美国ADC多光谱相机采集玉米拔节期冠层多光谱图像,利用德国AA3连续流动分析仪测定叶片氮含量。提取红色通道灰度均值(AVSR)、绿色通道灰度均值(AVSG)和近红外通道灰度均值(AVSNIR)等3个光谱参数,构建归一化植被指数(NDVI)、绿色归一化植被指数(GNDVI)、红色通道与绿色通道比值植被指数(RVIR/G)、红色通道与近红外通道比值植被指数(RVIR/NIR)、近红外通道与红色通道比值植被指数(RVINIR/R)、红色归一化比值(RNR)、绿色归一化比值(GNR)、近红外归一化比值(NIRNR)等8个植被指数。将全部光谱参数及植被指数分别与氮素值进行相关性分析,建立寒地玉米氮素一元线性回归、多项式回归及多元回归模型。结果表明:一元回归模型R2最高达0.854,多元回归模型R2为0.870,所得模型可为寒地大田玉米精准施肥和长势监测提供支持。

In order to rapidly acquire maize nutrient information in the field,a non-destructive method of maize nitrogen content index measurement was conducted based on multispectral imaging technique. Firstly,American ADC multi-spectral image monitoring system was available to acquire the canopy images of maize in jointing stage. At the same time,each sample was measured to show the nitrogen content index by AA3 continuous flow analyzer. Secondly,eleven vegetation indices were calculated including AVSR,AVSG,AVSNIR,NDVI,GNDVI,RVIR/G,RVIR/NIR,RVINIR/R,RNR,GNR and NIRNR. And then the method of correlation analysis was used to reduce the dimension of data so as to acquire three sensitive spectral characteristic parameters. Lastly,the nitrogen index detecting model based on simple linear regression method,polynomial regression method and multiple regression method by stepwise regression. The results indicated that,the maximal R2 of simple regression models is 0. 854 and the R2 of multiple regression model is 0. 870. It was feasible to diagnose nitrogen content of maize based on multi-spectral images.