一种红树林亚种分类的无人机高光谱遥感方法  

A UAV hyperspectral remote sensing method for mangrove subspecies classification

作　　者：孙玉鑫 梁庆炎 陈勇明 SUN Yuxin;LIANG Qingyan;CHEN Yongming(Surveying and Mapping Institute Lands and Resource Department of Guangdong Province, Guangzhou Guangdong 510670, China;Key Laboratory of Natural Resources Monitoring in Tropical and Subtropical Area of South China, Ministry of natural resources, Guangzhou Guangdong 510670, China)

机构地区：[1]广东省国土资源测绘院,广东广州510670 [2]自然资源部华南热带亚热带自然资源监测重点实验室,广东广州510670

出　　处：《北京测绘》2022年第6期762-766,共5页Beijing Surveying and Mapping

基　　金：广东省自然资源厅科技项目(GDZRZYKJ2020004);广东省自然资源厅科技项目(GDZRZYKJ—ZC2020003)。

摘　　要：利用大疆公司矩阵(Matrice)600专业型无人机搭载美国海德沃(Headwall)公司纳米高光谱(Nano-Hyperspec)型号微型高光谱成像仪获取南沙湿地公园一期范围的高光谱影像,配合已有高分辨率可见光影像和同步采集的试验区样本数据,基于像元解混和支持向量机对影像进行分类。实验表明:红树林亚种分类对样本质量和高光谱影像质量依赖度高;像元解混方法和支持向量机方法表现良好,整体分类精度均达到80%以上,其中像元解混方法精度可以达到83.93%;相比于支持向量机分类方法,解混方法能够在一定程度上提高红树林树种的分类精度。In this paper,a DJI Matrice 600 professional drone equipped with a Headwall Nano-Hyperspec miniature hyperspectral imageris in use to obtain hyperspectral images of the first area of the Nansha Wetland Park.There were high-resolution optical images and synchronously collected sample data from the experimental area,andhyperspectral images were classified based on pixel unmixing and support vector machines.Experiments showed that the classification of mangrove subspecies highly depended on the quality ofsampledata and hyperspectral image;both the pixel unmixing method and the support vector machine method performed well,and the overall classification accuracy was above 80%,while the pixel unmixing method had a better accuracy of 83.93%.Compared with the support vector machine classification method,the unmixing method could improve the classification accuracy of mangrove tree species to a certain level.

关 键 词：高光谱 红树林亚种 端元提取 像元解混 

分 类 号：P237[天文地球—摄影测量与遥感]