基于面向对象的GF-2卫星影像湿地信息提取  

Wetland information extraction of GF-2 satellite images based on object-oriented method

作　　者：孙祥龙 李卫萍 丁明君 王玉明 孙鹏 SUN Xianglong;LI Weiping;DING Mingjun;WANG Yuming;SUN Peng(Hangzhou Zhenshan Information Technology Co.,Ltd.,Hangzhou,Zhejiang 310012,China;Rizhao Natural Resources and Planning Bureau,Rizhao,Shandong 276800,China)

机构地区：[1]杭州臻善信息技术有限公司,浙江杭州310012 [2]日照市自然资源和规划局,山东日照276800

出　　处：《测绘标准化》2025年第1期139-143,共5页Standardization of Surveying and Mapping

摘　　要：传统的湿地信息提取主要采用外业调查和目视解译方法,存在耗时、费力且成本高的问题。近年来,利用遥感影像进行湿地信息提取成为研究的热点,但低分辨率遥感影像精度难以满足实际需求,高分辨率影像往往成本较高。本文基于高分二号(GF-2)卫星影像数据,建立湿地类型分类体系,采用面向对象分类方法提取南京市长江区域的湿地信息。结果表明,面向对象分类方法对研究区GF-2影像进行湿地信息提取的总体精度为90.86%,Kappa系数为0.904,分类精度较高,能够满足变化监测的要求,该研究结果可为利用GF-2卫星影像精准监测湿地变化提供支持。The traditional wetland information extraction mainly use field surveys and visual interpretation,it has the problems of time-consuming,labor-intensive,and high cost.In recent years,using remote sensing images to extract wetland information has become a hot topic,while the accuracy from low-resolution remote sensing images can't meet the practical requirements,and high-resolution images are typically expensive.Therefore,this paper constructs a wetland classification system based on Gaofen-2(GF-2)satellite images,and uses object-oriented classification method to extract wetland information in the Yangtze River region in Nanjing City.The results show that the object-oriented classification method can get good accuracy when extracing wetland information from GF-2 satellite images,the overall accuracy is 90.86%,and the Kappa coefficient is 0.904,which can meet the change monitoring requirements.This study can provide great support for accurate monitoring of wetland changes using GF-2 satellite images.

关 键 词：GF-2卫星影像 湿地 面向对象 信息提取 

分 类 号：P237[天文地球—摄影测量与遥感]