摘要
随着中国城市化进程的持续推进,城市内部空间结构不断变化。城中村作为城市内部结构的重要组成部分,在城市发展与环境建设中起着关键作用。因此,快速、准确地提取城中村建筑对城市研究具有重要意义。该文提出一种耦合多源大数据的方法提取城中村建筑物。首先,融合腾讯用户密度(TUD)数据、建筑物轮廓数据及兴趣点(POI)数据,利用基于密度的方法识别住宅区建筑物;其次,从住宅区中选取一定比例的建筑物作为训练样本,计算其占地面积、楼层高度、建筑密度和楼房间距信息,通过随机森林方法对上述特征进行训练以提取城中村。选取广州市天河区作为研究区,将城中村识别结果与实地调研数据、百度街景地图进行对比验证分析。结果显示:城中村的识别正确率超过89.29%;楼层高度及楼层间距对城中村建筑物提取的准确度起着关键作用;天河区的城中村呈现"外分散,内紧密"特点。研究证明,耦合多源大数据并采用基于密度和随机森林的方法识别城中村是可行且有效的。
China has been experiencing rapid urbanization at an unprecedented rate,urban internal spatial structure has changed significantly.Urban villages,resultant of urbanization,play a key role in urban development and environmental construction.Therefore,the extraction of buildings of urban villages is helpful not only for understanding the internal structure of the city,but also for establishing a basis of reconstruction of urban villages.Firstly,this paper integrated Tencent user density(TUD)data,building footprint data and points of interest(POI)data to identify residential buildings by using a density-based method in Tianhe District,Guangzhou,China.Secondly,partial residential buildings were selected as training samples to obtain features,including building area,floors,building density and the spacing of the buildings.The random forest analysis was applied to recognize buildings of urban villages.Finally,Baidu Street View Map and field surveys were used to verify the classification results.The study shows:1)the recognition accuracy of urban villages remains above 89.29%;2)the characteristics of building height and floor spacing have a great influence on the results;3)as a whole,the villages in Tianhe District distribute denser at the center.The results reveal that it is feasible and valid to identify urban villages adopting a density-based method and random forest analysis based on multi-source data.
作者
赵云涵
陈刚强
陈广亮
刘小平
牛宁
ZHAO Yun-han;CHEN Gang-qiang;CHEN Guang-liang;LIU Xiao-ping;NIU Ning(School of Geography and Planning,Sun Yat-Sen University,Guangzhou 510275;School of Economics and Management,Changsha University,Changsha 410022;Guangzhou Lantu Geography Information Technology Co.,Ltd.,Guangzhou 510663,China)
出处
《地理与地理信息科学》
CSCD
北大核心
2018年第5期7-13,共7页
Geography and Geo-Information Science
基金
湖南省自然科学基金项目(2016JJ6006)
关键词
城中村
建筑功能识别
多源大数据
随机森林
urban villages
building function identification
multi-source data
random forest
