【摘要】 近年来,随着信息技术的迅猛发展,高光谱遥感作为一种快速监测手段已经被广泛应用于农业生产中,利用高光谱遥感技术对农作物进行分类是一种无损、快速的技术手段。研究利用FieldSpec HandHeld 2TM手持式野外光谱辐射仪实地采集了大田油菜(Brassica napus L.)的苗期、抽薹期、花期样本光谱反射率数据;通过引入内聚系数来控制波段聚合性,利用特征波长指数来寻找识别特征波段,使用主成分分析法降维后用马氏距离的方法建立了识别模型。结果表明,不同品种油菜的光谱反射率样本存在差异,利用光谱数据识别区分不同品种油菜的适宜期为抽薹期,模型识别精度均达到了75%以上。表明高光谱遥感技术在不同品种油菜的识别上具有一定的应用优势。更多还原

【Abstract】 With the rapid advances of information technology, Hyperspectral Remote Sensing as a rapid means of monitoring has been used widely in agriculture, which is non-destructive for crop classification. The spectral reflectance data of three rapeseed oil varieties(Brassica napus L.) during the stage of seedling, bolting and flowering were collected by FieldSpec HandHeld 2TMin the fields. In order to set up the identification model of rape seed, the quality of data was controlled by Inner clustering coefficient(Inner-cc), identification bands were selected by the Characteristic Wavelength Index(CWI), and finally the Markov distance was determined by the data with dimension reduction through principal component analysis. Results showed that there were differences among spectral reflectivities of tested rapeseed, the accuracy of the identification model was no less than 75%, and the stage of bolting could be an appropriate stage for identifying rape seed varieties. It was suggested that the Hyperspectral Remote Sensing has good prospects in identifying rape seed varieties.更多还原