Analysis And Research On Hyperspectral Remote Sensing Characteristics Of Typical Crops In Central Yunnan

Spectral characteristics of surface features were the basis of remote sensing technology research and application,which reflected the characteristics,changes and differences of surface features.Hyperspectral remote sensing technology,because of its large number of bands and nano-scale spectral resolution,can capture tiny spectral differencesa and more accurately.At present,precision poverty alleviation has come to a perfect end,rural revitalization is being comprehensively promoted,and hyperspectral remote sensing will continue to play a significant advantage in precision agriculture services.In the dynamic supervision of agriculture,hyperspectral remote sensing technology can improve the level of agricultural informatization,and provide theoretical basis and technical support for monitoring the growth of typical crops in central Yunnan province and promoting the development of small plot and decentralized plateau characteristic agriculture.In this study,seven typical crops(cabbage,Yunnan mustard,white turnip,potato,wheat,corn and rape)in central Yunnan province(Kunming,Yuxi,Qujing and Chuxiong Yi Minority Autonomous Prefecture)were taken as research objects,and through field testthe,ground hyperspectral data(SOC 710vp)and satellite image data(GF-5)were obtained as data sources.Based on the original spectrum,using a variety of spectral transformation methods,extracting characteristic parameters for qualitative and quantitative analysis,and comprehensively selecting the spectral characteristic bands of crops to analyze the differences and similarities among crop species.The main results as the follows:1.In this paper,ENVI were used to preprocess the hyperspectral data of GF-5.Through a series of processing,the radiation and geometric distortion of the image were corrected,and the true reflectance of the ground object was restored.The result is good.2.Analysis of four kinds of full-band spectra(including original,first derivative,continuum removal and inverse-log spectrum)found that in the visible range of the original spectrum,radish and wheat had the highest discrimination.Derivative reflectance method highlighted slope of the spectral curve,can distinguish the spectral characteristics of white turnip and corn.By performing the cotinuum removal method,the absorption characteristics of spectral curves can be magnified,thus extracted seven"absorption valley"could distinguish different kinds of crops.Inverse-log spectrum could magnify the spectral difference of crops in the visible range and realize the transformation from"two valleys and one peak"to"two peaks and one valley".3.Two types of spectral feature variables(three-edge parameters and absorption feature parameters)were extracted for quantitative analysis.The results show that:Yunnan mustard's red and blue peaks were the largest.Seven crops may be affected by environmental factors,all of which had different degrees of blue shift,especially cabbage.The area ratio parameter(S_r/S_y)had the best effect in distinguishing seven crops.Among the absorption feature variables,the second absorption valley(550?760nm)had the most significant absorption intensity,and the width-depth ratio(WD)can amplify the difference.4.Based on spectral characteristic method and statistical analysis method(mean confidence interval method),the spectral characteristic bands of seven crops were concentrated in the green part of photosynthesis,the red edge with the fastest change of spectral reflectance and spectral high-reflection region in near infrared region.At the same time,the feature band selection achieves the effect of spectral dimension reduction.