伊犁绢蒿荒漠草地主要植物的高光谱识别

Hyperspectral Identification of Main Plants in the Desert Grassland of Seriphidium transiliense

以伊犁绢蒿荒漠草地为研究对象,利用SOC710 VP成像光谱仪获取不同季节的伊犁绢蒿荒漠草地地面高光谱影像,通过主成分分析和方差分析提取不同季节主要物种伊犁绢蒿、角果藜、叉毛蓬和裸地光谱特征波段、特征参数与植被指数;采用Fisher判别分析法将样本按7∶3的比例划分训练集和测试集,对3种主要植物和裸地进行识别。研究表明:6月植被指数建立的识别模型对测试集样本伊犁绢蒿、角果藜、叉毛蓬和裸地的判别精度最优,分别达到100%、95.56%、100%和100%,总体精度达98.89%;基于高光谱影像数据,以6月植被指数作为识别参数可以实现伊犁绢蒿荒漠草地主要植物的识别,为进一步对群落影像进行物种定量分类提供了依据。

In this study,we took the desert grassland of Seriphidium transiliense as the research object,obtaining ground hyperspectral images of desert grassland of Seriphidium transiliense in different seasons by using the SOC710 VP imaging spectrometer.Principal component analysis and variance analysis were used to extract main species Seriphidium transiliense,Ceratocarpus arenarius,Petrosimonia sibirica and bare land spectral characteristic bands,characteristic parameters and vegetation indices in different seasons.The Fisher discriminant analysis was used to divide the samples into training set and test set at a ratio of 7∶3,and the three main plants and bare land were identified.The results showed that the identification model established by the vegetation index in June had the best discrimination accuracy for the samples of Seriphidium transiliense,Ceratocarpus arenarius,Petrosimonia sibirica and bare land,reaching 100%,95.56%,100%,and 100%respectively,and the overall accuracy was 98.89%.Based on the hyperspectral image data,the main plants in the desert grassland of Seriphidium transiliense can be identified by using June vegetation index as identification parameter,which provides a basis for further quantitative classification of species in community images.