Study On Nutrition Diagnosis Of Nitrogen Content In Maize Leaves Of Cold Region Based On Hyperspectral Imaging

Maize is one of the main food crops in China,which plays an important role in the national economy.Nitrogen is one of the essential elements for plant growth.The reasonable application of nitrogen fertilizer can not only guarantee the basic nutrients needed by plants,but also increase the output,reduce the cost and protect the environment.Compared with the traditional diagnostic methods of chemical nitrogen,hyper spectral technology has high efficiency,nondestructive,and environmental protection,has become an important means to obtain nutrients,detection of crop growth.Prediction models of maize leaf N content using hyper spectral imaging technology,can provide guidance for maize plants of precision nitrogen.The test area of this study is located in city of Harbin Hei Longjiang province Fang Zheng County De Shan Xiang maize field,the research object is the jade 20 varieties,four important stages: test time selection for corn seedling,jointing,booting and filling stage.By the Headwall spectral imaging hyper spectral image acquisition instrument indoor platform of maize leaves at the same time,through the German AA3 continuous flow of ammonium nitrogen in leaf analyzer.Using the ENVI software to extract the original spectral reflectance data R in the picture,the original spectral data for five kinds of transformation processing methods: first derivative'R,second derivative R'',reciprocal 1/R,logarithm lgR,square root R.The original data and the 5 kinds of transform data and leaf nitrogen content were analyzed,obtained for the optimal wavelength in different growth stages,the optimal wavelength data and leaf nitrogen content by multiple stepwise regression analysis showed that the sensitive wavelength,dimensionality reduction of hyperspectral data.Through the analysis of the spectral curve,based on the location,area of the construction of high spectral parameters,ratio of difference,first derivative and two derivative based on sensitive wavelength data,high spectral parameters normalized,and the spectral parameters and nitrogen data correlation analysis of selected characteristic spectral parameters.The multiple linear regression model was established by using the spectral data and the N content of the leaf.The single variable regression model and multivariate linear model for the four growth periods were established by using the characteristic hyperspectral parameters;Through analysis,the establishment of four nitrogen content in growth period were the best prediction model of multivariate linear regression model of the sensitive wavelength data,based on the grain filling stage,jointing stage,booting stage,the best modeling accuracy at seedling stage decreased,high precision determination coefficient were 0.895,0.871,0.813,0.675.The prediction accuracy is 0.916,0.780,0.819,0.641.The accuracy of the model is high,which can provide a theoretical basis for the prediction of nitrogen content in maize leaves.