In order to explore the identification method and technology of crop varieties,the field Spec3 Hi-Res surface feature spectrometer was used in this study to collect the canopy spectral data of 11 rapeseed varieties at the pre-seedling,late seedling and bolting stages.Four data processing methods were used to analyze the canopy hyperspectral differences of rapeseed varieties at different growth stages.The position,amplitude,area,width,reflectivity and vegetation index of hyperspectral data were used as the research indexes,and the contribution rate of 23 characteristic parameters was used as the measurement index.The variance analysis method was used to identify the rapeseed varieties.It provides a basis for further study on the classification and identification of rapeseed varieties and regional management of crop planting.The main research results are as follows:1.In view of the characteristic differences of the hyperspectral curves of different rapeseed varieties,four data processing methods:mean,first-order differential,normalization transformation and logarithmic transformation were used to study the difference of the canopy hyperspectral curves of each variety in different periods.The research results show that among different data processing methods,the processing effect of mean value processing and logarithmic transformation is the best.There were significant differences in the reflectance spectra of 11 rapeseed varieties,with the most obvious difference in the 760~900 nm band,followed by the 520~680 nm band.2.Aiming at the effect of canopy hyperspectral characteristic parameters on the identifying rapeseed varieties,the contribution rate of characteristic parameters was used to explore the ability of six types of characteristic parameters to identify rapeseed varieties.The results showed that different varieties of rapeseed have differences in characteristic parameters.The identification effects of the same type of characteristic parameters are different in different periods.Two types of area and reflectivity characteristic parameters have the best recognition effect at bolting stage,three types of position,amplitude,and width characteristic parameters have the best recognition effect at early seedling stage,vegetation index parameters had the best recognition effect at late seedling stage.The characteristic parameters in the same period have different ability to distinguish rapeseed varieties.In the early seedling stage,the amplitude parameter had the best recognition effect,followed by the reflectivity parameter.In the late seedling stage,the amplitude parameter identification effect is the best,followed by the area parameter.At bolting stage,area parameter identification effect is the best,followed by amplitude parameter.The comprehensive effect from strong to weak is:amplitude parameter>area parameter>reflectivity parameter>vegetation index parameter>position parameter>width parameter.3.For the optimal characteristic parameters for screening and identifying rapeseed varieties,one-way analysis of variance was used,the results showed that the comprehensive effect of identifying varieties with four characteristic parameters of D_r,SD_r,SD_r/SD_b,(SD_r-SD_y)/(SD_r+SD_y)was good.The Waller-Duncan multiple comparison analysis of variance was used to compare the four parameters.,the results show that the amplitude parameter D_r has the best recognition effect.Among them,in the late seedling stage,three characteristic parameters D_r,SD_r/SD_b,(SD_r-SD_y)/(SD_r+SD_y)have the best recognition effect,which can clearly distinguish six varieties.In the late seedling stage,four characteristic parameters of D_r,D_b,(ρ_g-ρ_r)/(ρ_g+ρ_r)and SD_y were the best,which could clearly distinguish eight varieties.4.Aiming at the best period for identifying rape varieties,the above three points were comprehensively analyzed,namely,the results of hyperspectral curve,contribution rate of characteristic parameters and variance analysis at the early seedling stage,late seedling stage and bolting stage.The results showed that bolting stage was the optimal identification period of rapeseed varieties.The main highlights of this study.Firstly,it is found that there are differences in canopy hyperspectral curves among different varieties of rapeseed.The band of 760~900 nm is the most obvious,followed by the band of 520~680 nm.The above bands can be used to identify rapeseed varieties.Secondly,there are differences in characteristic parameters among different varieties of rapeseed.Rape varieties can be identified by the contribution rate of characteristic parameters,and the identification effect of amplitude and area characteristic parameters is the best.In terms of the growth period of rapeseed,there were extremely significant differences among different characteristic parameters at bolting stage,and it was considered that bolting stage was the best time to identify rapeseed varieties.In terms of characteristic parameters,characteristic parameters D_r、SD_r/SD_b、(SD_r-SD_y)/(SD_r+SD_y)had the best recognition effect at seedling stage,and characteristic parameters D_r、D_b、(ρ_g-ρ_r)/(ρ_g+ρ_r)、SD_yhad the best recognition effect at bolting stage. |