论文部分内容阅读
为了研究适用于湿地植物分类的高光谱植被指数及其分类效果,以三江平原洪河国家级自然保护区为研究区,使用HR-1024光谱仪,获取了研究区内毛薹草(Carex lasiocarpa)、小叶章(Calamagrostis angustifolia)、乌拉草(Carex meyeriana)、漂筏薹草(Carex pseudocuraica)和大豆(Glycine max)5种植物的冠层光谱数据,对获取的高光谱数据进行去水汽、平滑处理,以消除环境背景的影响;总结了前人文献中使用的38种高光谱植被指数,研究这些高光谱植被指数对研究区内典型植物分类的的适用性;然后,采用方差分析方法,从所有高光谱植被指数中筛选出区分频率较高的25种指数;最后,采用Fisher线性判别分析法,进行植物类型的识别,并对分类结果进行精度评价与分析。研究结果表明,利用高光谱植被指数进行湿地植物分类的精度总体为85%,高光谱植被指数可以作为植物分类的依据。
In order to study the Hyperspectral Vegetation Index and its classification effect which are suitable for wetland plant classification, taking Honghe National Nature Reserve of Sanjiang Plain as study area, the Carex lasiocarpa, The data of canopy spectra of five species including Calamagrostis angustifolia, Carex meyeriana, Carex pseudocuraica and Glycine max were de-watering and smoothing the obtained hyperspectral data, To eliminate the impact of the environmental background; 38 kinds of hyperspectral vegetation indices used in the previous literature were summarized, and the applicability of these hyperspectral vegetation indices to the classification of typical plant species in the study area was studied; and then, from variance analysis, Spectral index of vegetation were screened out a higher frequency of 25 kinds of index; Finally, using Fisher linear discriminant analysis method for plant type identification, and the classification results of the accuracy evaluation and analysis. The results showed that the accuracy of classification of wetland plants by hyperspectral vegetation index was 85%, and hyperspectral vegetation index could be used as basis for plant classification.