Study On The Spectral Preprocessing Method And Water Detection Model Of Steed Jujube Leaves

Leaf water content is an index of the growth status of jujube trees.A series of structural forms of leaf color and growth status of jujube will change with the moisture content of jujube leaves.Therefore,it can be detected and diagnosed in real time according to the water content of leaves.Near infrared spectroscopy(NIR)technology is characterized by its convenience,rapidity,accuracy,and no destructiveness.It can quickly collect relevant data of Jun jujube leaves,and has been the focus of research in recent years by using NIR spectroscopy to detect leaf water content.In this study,5 years in 10 group 6 with Jun jujube leaves as experimental objects,the use of portable near infrared spectrometer(sup NIR-15201000-1800nm)collected 121 pieces of jujube leaf spectra,the original spectra of Junzao images using MC(spectral mean center),autoscaling(standard),normalization(normalized),MSC(multiple scattering correction),SNV(standard normal variate)and WT(wavelet transform)six pretreatment method in Window7 environment,processing spectral image by Matlab R2014 a software,build the measurement model of moisture content of Jun jujube leaf by PLS,and the evaluation of model parameters are compared to the moisture content of Jun jujube leaf the fast and non destructive testing provide the basis for.The main results are as follows:(1)investigated the effects of different pretreatment methods on the model,R,RMSEPCV,precision,RMSEP as the evaluation model of the merits of the basis,through the comparison,through the spectral model is better than the effect of modeling after pretreatment of the untreated,MSC and SNV two pretreatment established a good mathematical model.(2)in the wavelength selection,screening jujube leaf water by SPA method of characteristic wavelength,MSC by SPA algorithm to select 5 optimal wavelengths were 1719 nm,1773nm,1011 nm,1427nm,1582 nm,1011nm in considering the need to eliminate edge;SNV by SPA algorithm to select 5 optimal wavelengths,respectively,1719 nm 1799nm,1126 nm,1429nm,and1660 nm.The results show that the water model of steed jujube leaves can meet the needs of the actual production.(3)excluding the abnormal samples before dividing the calibration set samples,according to the cumulative contribution rate of principal components,4 principal components were selected from the leaves of Jun jujube.4 samples were removed from Maozao leaves by Mahalanobis distance,and the sample numbers were 40,49,68 and 78.The SPXY method was used to carry out the branch of the branch of Jun jujube after removing the abnormal samples.(4)the characteristic spectral bands of SPA extracted by using partial least squares modeling,MSC treated R(predicted correlation coefficient)increased from 0.6204 to 0.7504,RMSEP(root mean square error of prediction)decreased from 0.0407 to 0.0343,Precision(precision)increased from 0.9438 to 0.9510;after SNV treatment,R(correlation coefficient forecast)increased from0.6204 to 0.7089,RMSEP(root mean square error of prediction)decreased from 0.0407 to 0.0327,Precision(precision)increased from 0.9438 to 0.9577.