基于GF-2卫星影像的连云港滨海湿地植被类型分类
Vegetation Classification of Lianyungang Coastal Wetland Based on GF-2 Satellite Images
高雨 1胡召玲 1王圳 2高亚军 2许存培 3顾洁 4陈佩龙 5朱孝天5
作者信息
- 1. 江苏师范大学 地理测绘与城乡规划学院,江苏 徐州 221116
- 2. 连云港林业技术指导站,江苏 连云港 222000
- 3. 连云港市不动产交易登记中心,江苏 连云港 222006
- 4. 连云港市国土资源信息中心,江苏 连云港 222001
- 5. 江苏海洋大学 海洋技术与测绘学院,江苏 连云港 222005
- 折叠
摘要
选择连云港海岸带临洪河口湿地和埒子河口湿地为研究区,以GF-2 号卫星影像为数据源,经几何校正和大气校正后,使用Gram-Schmidt算法进行影像融合;利用融合后影像基于面向对象方法,分别采用随机森林和最邻近法进行影像分类.实验结果表明:GF-2 号卫星较好地反映了不同时相滨海湿地植被的光谱特征和空间特征,使用随机森林算法结合光谱和空间结构特征能较好地对滨海湿地植被类型进行分类;影像的时相特征和空间结构明显地影响植被分类精度.
Abstract
This paper selects the Linhong estuary wetland and Liezi estuary wetland of Lianyungang coastal zone as the research area.Using GF-2 satellite images as data source,after geometric correction and atmospheric correction,image fusion is performed by Gram-Schmidt algorithm;based on object-oriented method,random forest and nearest neighbor method are used to classify fusion images.The results show that GF-2 satellite images can better reflect the spectral and spatial characteristics of coastal wetland vegetation in different phases,and the vegetation types of coastal wetland can be well classified by using random forest algorithm combined with spectral and spatial structure characteristics.The temporal characteristics and spatial structure of images obviously affect the accuracy of vegetation classification.
关键词
滨海湿地/植被类型/GF-2号卫星影像/随机森林/面向对象Key words
coastal wetlands/vegetation types/GF-2 satellite images/random forest/object-oriented引用本文复制引用
出版年
2024
NETL
NSTL
万方数据