Research On Hyperspectral Detected Method Of Soil Total Nitrogen

Soil total nitrogen is an important indicator of soil fertility and plays a crucial role in crop growth.The hyperspectral spectrum has the characteristics of high resolution,continuous band and uniform pattern,and can quickly and efficiently detect total nitrogen content in soil.Due to the large amount of hyperspectral information and complex data,it can easily lead to collinearity and redundancy of spectral data,which seriously affect the effect of hypothermia prediction model of soil total nitrogen.Therefore,how to solve the existence of collinearity and redundancy of spectral data,to build high-precision soil total nitrogen hyperspectral detecting model is an urgent problem which needed to be solved.In this study,based on the low accuracy of the hyperspectral technique to detect soil nitrogen content,the soil nitrogen of Shajiang black soil in northern Anhui as the research object,focusing on the two key in the construction of soil total nitrogen hyperspectral detection model—the method of extracting the characteristic wavelength of soil total nitrogen and the optimization problem in the method of building hyperspectral detection of soil total nitrogen.The method of extracting spectral characteristic wavelength of soil total nitrogen and the method of hyperspectral detection of soil total nitrogen is studied.A hyperspectral data analysis system for soil total nitrogen was developed.The main contents and achievements of the paper are as follows:(1)The extraction method of spectral characteristic wavelength of soil total nitrogen was studied.The competitive adaptive weighting algorithm,continuous projection algorithm and random forest algorithm were used to select the characteristic wavelength of soil total nitrogen.And the soil total nitrogen hyperspectral detection model was established by the characteristic wavelength.The selected characteristic wavelength was required to establish the highest model with the lowest number of wavelengths.(2)The optimal model of hypothermia detection of soil total nitrogen was studied.On the basis of the extraction of the characteristic wavelength,the Partial Least Squares method,the Support Vector Machine regression and the Lasso Regression model were used to detect total nitrogen content in soil,and the mapping relationship between the spectral and soil total nitrogen content was established and Soil total nitrogen content was predicted by mapping relation.Compared with the above three modeling methods,the Lassoregression model is found as the model with the highest prediction accuracy,which can accurately detect the content of nitrogen in soil and lay the foundation for reasonable fertilization.(3)soil total nitrogen hyperspectral data analysis system was developed.In the Eclipse platform using IDL,Java,MATAB language,we completed the development of the soil total nitrogen hyperspectral data analysis system,the system had solved he collaborative work of many software in the processing of data dependence,the software data format is not compatible,the loss of precision in the process of data transfer and other issues.The hyperspectral data of this study were analyzed and validated by this system.The results of this paper have important theoretical and practical significance for the construction of soil total nitrogen hyperspectral fast non-destructive testing model,reasonable fertilization,protection of the environment and improvement of crop production.