论文部分内容阅读
红树林种类识别对于研究红树林生态系统的变化具有重要意义。本文以广西铁山港红树林区为研究区域,以国产资源三号测绘卫星数据为数据源,分析区内各种红树林的光谱特性,并结合多项植被指数(RVI,NDVI,VARI和NDGI)信息,采用联合稀疏表示分类器进行红树林种类遥感识别。本文主要分析了桐花树、海漆、白骨壤、红海榄、秋茄、海桑、木榄这7种红树林种类以及陆地灌丛、泥滩、草地这些非红树林种类的几项植被指数,并结合多光谱图像的几何空间与光谱特征空间,采用联合稀疏表示算法进行红树林种类分类。利用组合光谱和4种植被指数信息进行分类可以达到最好的分类效果,总体精度为95.37%,kappa系数为0.9347。实验结果表明:光谱特征结合植被指数信息进行分类能提高分类精度,四种植被指数中NDVI对于区分红树林种类具有更大的贡献,联合稀疏表示分类器在红树林种类识别中表现出优异的分类效果。
Mangrove species identification for the study of mangrove ecosystem changes is of great significance. In this paper, the mangrove area of Guangxi Tieshangang was taken as the research area. The spectral characteristics of mangrove forests in the area were analyzed with the data of domestic resource No.3 mapping satellite, and combined with several vegetation indices (RVI, NDVI, VARI and NDGI ) Information, a joint sparse representation classifier for mangrove species remote sensing identification. In this paper, the author mainly analyzed seven kinds of mangrove species including Aegiceras corniculatum, Sea lacquer, Avicennia marina, Rhizophora stylosa, Candel, Haisang and Mu-lam and several vegetation indices of non-mangrove species such as land shrubs, mudflats and grasslands, Combined with the geometric space and spectral feature space of multispectral image, a joint sparse representation algorithm is used to classify mangrove species. The best classification results were achieved using the combined spectra and the four vegetation index information, with an overall accuracy of 95.37% and a kappa coefficient of 0.9347. The experimental results show that the classification accuracy can be improved by combining spectral features and vegetation index information. NDVI of four vegetation indices has a greater contribution to the classification of mangrove species, and the joint sparse representation classifier shows excellent classification in mangrove species identification effect.