摘要
针对多源数据提取建成区的研究大多集中在等权融合方向,缺乏考虑不同数据源包含的信息量差异的问题,通过多源数据加权融合方法对前人的城市建成区提取指数进行改进,以改善不同质量级数据直接融合导致建成区提取不准确的问题。首先,构建融合夜间灯光数据、NDVI数据、路网数据和POI数据的改进PREANI指数;其次,融入不透水面数据和温度数据构建无权指数UCI和有权指数WCI;最后,选择迭代法和动态阈值法提取建成区,并对3种指数分别进行评估。结果表明:改进PREANI的Kappa系数为0.80,UCI融入温度和不透水面数据将Kappa系数提高至0.83;WCI提取的建成区轮廓更准确,在增强城乡建成区细部对比、提升边缘地物区分能力方面表现良好,其Kappa系数、查全率、查准率和F1分数均在0.85以上。
The research on multi-source data extraction in built-up areas predominantly emphasizes equal-weight fusion,overlooking the variations in information content among diverse data sources.The paper proposes an improvement to the urban built-up area extraction index from previous studies through a multi-source data weighted fusion method.The aim is to address the issue of inaccurate built-up area extraction resulting from the direct fusion of data with different quality levels.Firstly,construct an improved PREANI index by integrating nighttime light data,NDVI,road network data,and POI.Then,incorporate impervious surface data and temperature data to formulate both the unweighted index(UCI)and the weighted index(WCI).Finally,employ the iterative method and dynamic threshold method to extract the built-up area and evaluate the three indices separately.The results show that:the Kappa coefficient of PREANI has improved to 0.80,and UCI,with the incorporation of temperature and impervious surface data,increases the Kappa coefficient to 0.83;the constructed area contour extracted by WCI is more accurate and performs well in enhancing the detailed comparison of urban and rural built-up areas and improving the ability to distinguish edge features,and its Kappa coefficient,recall,precision,and F1 score are all above 0.85.
作者
马洋
牟凤云
左丽君
邵志豪
邹昕宸
MA Yang;MU Fengyun;ZUO Lijun;SHAO Zhihao;ZOU Xinchen(College of Smart City,Chongqing Jiaotong University,Chongqing 402247,China;Aerospace Information Research Institute,Chinese Academy of Sciences,Beijing 100101,China)
出处
《遥感信息》
CSCD
北大核心
2024年第2期110-117,共8页
Remote Sensing Information
基金
国家自然科学基金(T2261129473)
重庆交通大学研究生科研创新项目(2023S0129)。
关键词
建成区提取
多源数据加权融合
夜间灯光
路网
POI
不透水面
built-up area extraction
multi-source data weighted fusion
NTL
road network
POI
impervious surface
