Extraction And Monitoring Of The High Spectral Characteristics Of Winter Wheat In Winter Wheat Based On Wavelet Transform

This test by winter wheat(Linmai 7006)as the research object,using the refrigeration equipment to freeze injury stress test of winter wheat jointing stage,five days,ten days after the determination of frozen,spectral reflectance after 20 days and 35 days and agronomy parameter,analyzes the change law of wavelet transform processing method to deal with canopy spectra and extract the wavelet energy coefficient,using partial least squares(PLSR)analysis method to build canopy spectra and quantitative monitoring model of main agronomic parameters.The results showed that:1 The variation trend of the coronal spectra curve of winter wheat was basically consistent after different freezing treatment.In the early stage of frostbite,visible band changes are not obvious,and the reflectance of near-infrared band is greatly improved.However,over time,the overall change trend of the near infrared region band is very large,and the phenomenon of red valley uplift occurs.In the late period of freeze injury,the trend of visible band changes tends to be gentle.Although the near infrared band has a large change trend,the similarity of the overall trend gradually increases,and the red valley is not obvious,and the green peak is weakened.In addition,the phenomenon of "red border and blue shift"appeared in the spectral curve of the canopy after the treatment of frostbite.2 After the treatment of different freezing conditions,the main agricultural parameters of winter wheat changed significantly and showed a decrease trend.The results showed that the difference between the experimental group and the control group decreased gradually with the increase of the degree of freeze injury stress and the extension of the number of days.3 The winter wheat canopy original spectrum and main agronomic parameters(SPAD,plant height,the ground biomass)build the model,the model precision of R2 is 0.551,0.874,0.777,root mean square error is 3.265,5.136,2.128,RPD was 1.500,2.834,1.837,and verify the model R2 is 0.469,0.834,0.702,root mean square error is 3.563,5.904,5.084,RPD was 1.373,2.465,2.863.It is shown that the original spectral model based on PLSR method is good.4 After the wavelet energy coefficient and the main agronomic parameters were constructed by 5 different wavelet basis transformation processes,the optimal wavelet basis was Sym4 wavelet base,Bior1.3 wavelet base and Biorl.3 wavelet basis.Moreover,the accuracy of the model and RPD were significantly improved,and the root mean square error decreased.It is shown that the wavelet transform can improve the precision of the model and increase the application value of the model.