The Evaluation Model Of Winter Wheat Chlorophyll High Spectral Characteristics Extraction Under Low Temperature Stress

Winter wheat is often encountered under low temperature condition in the process of growth,which directly leads to a series of problems,such as the reduction of wheat production and seriously affects the food security of our country.Because the winter wheat has the characteristics of concealment and delay after freezing injury,so it is difficult to identify the economic losses by the naked eye.In recent years,with the continuous development of hyperspectral technology,it is possible to realize the real-time monitoring of winter wheat damage by using spectral technology,which has very important significance on the production and disaster prevention in winter wheat.Through pot experiment and field experiment,this paper researched the jointing stage of Winter Wheat in different low temperature treatment,and analysed a variety of hyperspectral data pretreatment method to remove the influence of the background and noise;The spectra information mining and extraction of chlorophyll content of winter wheat by the introduction of SPA,PLSR and SMLR;The validation and optimization of spectrum monitoring model of winter wheat chlorophyll content,and comprehensive evaluated the spectral data processing and information mining method.Comparing the advantages and disadvantages of different pretreatment methods and characteristics of band extracting method,winter wheat chlorophyll content of early estimates and frost damage degree evaluation was realized.The results showed that:1.After the occurrence of low temperature stress,the chlorophyll content of winter wheat decreased with the increase of stress intensity.With the development of the growth period,the chlorophyll content of the plant was gradually increased.2,Normalize,MSC correction treatment did not significantly improve the correlation between the pretreatment spectrum and chlorophyll content,while Baseline and CR decreased the correlation between the variables,but by comparing the correlation coefficient of the pretreatment,The maximum or mutation value of the correlation coefficient of the canopy spectrum is in the same band position,and has certain research value.(Rc2>0.93,RMSEc>0.22,RPDc>4.413;Rv2>0.86,RMSEv>0.48,RPDv>2.05)were selected as the best spectal resolution bycomprehensive evaluation and selection of the first-order differential transformation approach.3.The spectral bands SMO,SM1,SM3 and SM4 of winter wheat were extracted by continuus projection algorithm(SPA),partial least squares regression(PLSR)and stepwise multiple linear regression(SMLR).The spectral bands were extracted from 400-700 nm(visible light band)And 700-1250 nm(near infrared band)were the sensitive band regions of chlorophyll content,and a strong reflection peak near 550 nm"green peak" indicated that the area could characterize the chlorophyll information of winter wheat after low temperature stress.4.The feature band extracted by SPA method was superior to PLSR-SMLR method,and the whole spectrum modeling was better than the characteristic band modeling method,and has a certain application potential.5.By using the full spectrum band of winter wheat chlorophyll content estimation model,a nonlinear SVM model was superior to linear PLSR model,linear PCR model was the worst;Comprehensive evaluation of the results showed that the SVM nonlinear model was higher precision,and had a certain robustness and universality.