Study On Hyperspectral Characteristics And Classification Recognition Model Of Ramie Leaves

In recent years,hyperspectral analysis technology has been playing an increasingly important role in agriculture,such as farmland information acquisition,crop growth judgment,crop yield estimation and so on.Ramie,as an important source of textile fiber,has always been in a higher position in the national economy.At present,there are few studies on the hyperspectral characteristics of ramie at home and abroad,and the in-depth study of the hyperspectral characteristics of ramie is beneficial to the development of ramie and germplasm resources,filling the blank of previous research.This paper discusses the ramie leaf spectral characteristics of the original data,peak parameters,vegetation index,three edge parameters,skewness and kurtosis parameters of Hyperspectral Feature Index for statistical analysis,to understand and compare the spectral characteristics of different genotypes of ramie,the difference between high spectral characteristics and with other crops,for the establishment and classification of ramie the hyperspectral identification model,to explore the correspondence between ramie hyperspectral characteristics and crop physiological and ecological parameters provide theoretical basis based on.A classification model of ramie varieties based on hyperspectral remote sensing was established by stepwise discriminant analysis.A total of 652 leaf hyperspectral data were collected from 4 ramie cultivars with different genotypes under field cultivation conditions.According to ramie leaf hyperspectral reflectance curve,the extraction method of 4 parameters: Based on original data based on peak valley parameters,vegetation index,skewness and kurtosis parameters were established based on multiple linear discriminant function Fisher 4 feature parameters based on stepwise discriminant method based on three parameters,and the obtained model are compared original data: the correct rate of recognition is 88.89% based on the wave trough;the correct rate based on vegetation index is 88.87%;the correct rate of three based on 79.89% parameters;the correct rate is 89.77% based on Skewness and kurtosis.The classification results of four methods of feature extraction based on the above point of view: using stepwise discriminant method for identification and classification of ramie varieties,the lowest three correct edge parameters based on the method of feature extraction method based on Skewness and kurtosis effect is the best.The identification model of brown spot disease of ramie was established by stepwise discriminant analysis.A total of 268 leaf hyperspectral data were collected under field cultivation conditions with brown spot disease and healthy ramie varieties.According to ramie leaf hyperspectral reflectance curve,the extraction method of 4 parameters: Based on original data based on peak valley parameters,vegetation index,skewness and kurtosis parameters were established based on multiple linear discriminant function Fisher 4 feature parameters based on stepwise discriminant method based on three parameters,and validate the model.Based on the original data,the correct rate of wave trough is 100%,the accuracy based on vegetation index is 91.7%,the correct rate based on three side parameters is 91.6%,and the correct rate based on Skewness kurtosis is 100%.The classification model of four methods of feature extraction based on the above point of view: three edge parameters resolution model is compared with other varieties between the lowest rate of correct classification: feature extraction method of skewness and kurtosis and original data of wave crests and troughs,is a method to identify the effect of.