基于GF-1 PMS的喀斯特城镇绿地提取研究  

Research on extracting green space in Karst towns based on GF-1 PMS

作　　者：王玮 黄林 任金铜 WANG Wei;HUANG Lin;REN Jintong(Anhui Yuntu Space Information Technology Co.,Ltd.,Hefei 230022,China;Guizhou University of Engineering Science,Bijie 551700,China;Guizhou Province Key Laboratory of Ecological Protection and Restoration of Typical Plateau Wetlands,Bijie 551700,China)

机构地区：[1]安徽云图空间信息科技有限公司,安徽合肥230022 [2]贵州工程应用技术学院,贵州毕节551700 [3]贵州省典型高原湿地生态保护与修复重点实验室,贵州毕节551700

出　　处：《江苏科技信息》2024年第12期137-140,共4页Jiangsu Science and Technology Information

基　　金：喀斯特高原资源与环境遥感人才团队,项目编号:毕委人领通〔2023〕14号;智慧地理空间信息应用工程中心,项目编号:毕科联合〔2023〕8号;贵州省区域内一流建设学科“生态学”,项目编号:黔教XKTJ〔2020〕22。

摘　　要：绿地面积作为园林城市建设的重要指标,对于城市规划建设具有重要意义。文章基于面向对象分类算法,以位于喀斯特山区的贵州省毕节市海子街镇为研究区,利用高分一号(GF-1)PMS遥感影像进行城镇绿地提取。结果表明:绿地信息在GF-1 PMS影像上具有明显的光谱、纹理特征,通过对影像对象特征的不同选择和组合,能够很好地提取城镇绿地信息;在面向对象分类算法中,基于规则的绿地信息提取总体分类精度为79.36%,Kappa系数为0.76;基于样本的绿地信息提取总体分类精度为89.79%,Kappa系数为0.85,其提取效果比基于规则的面向对象方法更好。Green space area as an important indicator of garden city construction is of great significance for urban planning and construction.Based on object-oriented classification algorithm,the article takes Haizijie town,Bijie city,Guizhou province,located in the Karst mountainous area as the research area.The GF-1 PMS remote sensing image is used to extract urban green spaces.The results show that green space information has obvious spectral and texture features on GF-1 PMS images.By selecting and combining different image object features,urban green space information can be effectively extracted;in object-oriented classification algorithms,the overall classification accuracy of rule-based green space information extraction is 79.36%,the Kappa coefficient is 0.76;the overall classification accuracy of green space information extraction based on samples is 89.79%,the Kappa coefficient is 0.85,and its extraction performance is better than rule-based object-oriented methods.

关 键 词：高分一号 城镇绿地 面向对象 信息提取 

分 类 号：TP7[自动化与计算机技术—检测技术与自动化装置]