The Establishment Of The Estimation Model For The Content Of Tieguanyin NPK Based On Hyperspectral

The quality of tea generally refers to color,aroma,taste and shape.It includes whole effect of the sensory stimulation of the human body after dissolving the many compounds in tea,is required attributes that meet people's needs,and is influenced by various components such as free amino acid,soluble sugar,tea polyphenols and catechin.Nitrogen,phosphorus and potassium are essential components of tea quality formation.Nitrogen that can make tea leaves fat and strong is the constituent of all proteins and amino acids,as well as the main components of chlorophyll,various hormones and nucleic acids.Phosphorus is helpful to improve the metabolism of growth of tea trees,can increase the content of polyphenolic compounds such as polyphenols,epicatechins,and gallicin,and has a good effect on improving the aroma and taste of tea.Potassium is beneficial to promote the absorption of nitrogen and improves the intensity of photosynthesis.Although Potassium is not a component of organic matter in tea trees,it takes part in various biochemical processes in the form of ions,and it exists in the organic body that affects the quality of tea.This study use statistical regression analysis and the analysis method that based on characteristic spectral position variable to make estimation research on the total nitrogen,total phosphorus and total potassium content in teas that from different places in Anxi county Fujian province under the guidance of theory hyperspectral remote sensing inversion of biochemical components of vegetation and extraction of agronomy parameters.The purpose is to find spectral sensitive bands and characteristic parameters related to the content of biochemical components in tea leaves,establish a relational model between nitrogen,phosphorus and potassium content and a relational model of spectral characteristic band.Besides this study has test and analyze the models.The main conclusions of this study include:(1)Processing of spectral data using derivative spectrum technique,SG smoothing,red edge displacement analysis and continuum removal,which correlated with biochemical component content of nitrogen,phosphorus and potassium is higher derivative spectroscopy technology and SG convolution smoothing,but the correlation between the spectra of the red edge displacement analysis and the continuous defractionation was low.(2)In establishing process of estimating model,the study select independent variables by using derivative spectroscopy technology,the SG convolution smoothing sensitive wave band,red edge displacement analysis and continuum,and select content of nitrogen,phosphorus and potassium in biochemical components as dependent variables to build model.It is concluded that the model fitting effect of spectral sensitive bands as independent variables is better,while the fitting effect of spectral characteristic parameters as independent variables is not ideal,which is not suitable for the estimation of nitrogen,phosphorus and potassium contents in biochemical components of Tieguanyin.(3)Description of not ideal result that model using spectral characteristic parameters as independent variables to estimate NPK content:The estimation of nitrogen,phosphorus and potassium content in tea biochemical components by spectral characteristic parameters may also need to be combined with spectral vegetation index and other physical parameters of tea varieties.There are other biochemical components in the estimation of leaf potential spectral characteristic parameters such as the "trilateral"parameters,green mountain,red valley and hyperspectral vegetation index.Wave band depth,deep center normalization and band area normalization variables also provide other biochemical components in inversion of vegetation with more choices.(4)Using single variable regression analysis,multivariate stepwise regression analysis and partial least-squares method,the estimation model of tieguanyin biochemical components was established.It is concluded that the fitting effect of multivariate regression analysis and partial least squares method is better than that of single variable regression analysis model,which indicates that the change of biochemical component content is seldom led by single variable,but is influenced by multiple variables.The model of the second order transform spectral variable is generally better than that of the first order transform spectral variable,which may be related to the spectral reflectivity difference of the selected sensitive band.(5)The optimal model for the estimation of nitrogen,phosphorus and potassium content of tieguanyin biochemical components is obtained by using partial least squares analysis method.The best sensitive bands selected by the two order derivative transform are mainly distributed in the near infrared(760nm-1 300nm)range.