测绘学报 ›› 2020, Vol. 49 ›› Issue (7): 907-920.doi: 10.11947/j.AGCS.2020.20190315

• 地图学与地理信息 • 上一篇    下一篇

顾及兴趣点潜在上下文关系的城市功能区识别

陈占龙, 周路林, 禹文豪, 吴亮, 谢忠   

  1. 中国地质大学(武汉)地理与信息工程学院, 湖北 武汉 430074
  • 收稿日期:2019-07-28 修回日期:2019-12-31 发布日期:2020-07-14
  • 通讯作者: 周路林 E-mail:zhoululin@cug.edu.cn
  • 作者简介:陈占龙(1980-),男,博士,副教授,研究方向为空间分析算法、空间推理、地理信息系统软件开发与应用。E-mail:chenzhanlong2005@126.com
  • 基金资助:
    国家自然科学基金(41871305);国家重点研发计划(2017YFC0602204)

Identification of the urban functional regions considering the potential context of interest points

CHEN Zhanlong, ZHOU Lulin, YU Wenhao, WU Liang, XIE Zhong   

  1. School of Geography and Information Engineering, China University of Geoscience, Wuhan 430074, Chinat
  • Received:2019-07-28 Revised:2019-12-31 Published:2020-07-14
  • Supported by:
    The National Natural Science Foundation of China(No. 41871305);The National Key Research and Development of China(No. 2017YFC0602204)

摘要: 城市功能结构的探索对人们理解城市及城市规划有着重要的作用。兴趣点(points-of-interest,POI)数据作为城市设施的代表,被广泛应用于城市功能区提取。以往对城市功能区研究大多只考虑了POI统计信息,忽略了POI中丰富的空间分布信息,而POI空间分布特征与区域功能密切相关。本文利用空间共位模式挖掘方法挖掘POI潜在上下文关系,提取POI空间分布信息,构建区域特征向量,并进行区域聚类;再利用POI类别比例、居民的出行特征等对聚类结果进行识别。以北京市核心城市功能区为例,将研究结果与北京市百度地图、居民出行特征进行对比验证分析。试验表明,本文方法能识别出具有明显特征的城市功能区,如成熟的娱乐商业区、科教文化区、居住区等。同时,与基于POI语义信息的LDA方法及顾及POI线性空间关系的Word2Vec方法进行对比分析,证明了本文方法的优越性。

关键词: 城市功能区识别, 上下文关系, 兴趣点, 空间共位模式, 北京

Abstract: The exploration of urban functional structure plays an important role in understanding urban and urban planning. POI(point of interest) data, as a representative of urban facilities, is widely used to extract urban functional areas. In the past, most of the researches on urban functional areas only considered POI statistical information. However, they ignored the abundant spatial distribution characteristics of POI, which are closely related to regional functions. Therefore, we firstly use spatial co-location pattern mining to mine the potential context of POI, extract the spatial distribution information of POI, construct regional feature vectors, and carries out the regional clustering through the clustering algorithm. Then we use the POI class ratio and residents’ travel characteristics to identify the clustering results. We experimented our method on the core urban functional areas of Beijing, the results, which were verified with Baidu Map and residents’ travel characteristics, showed that they can identify urban functional areas with obvious characteristics, such as mature entertainment business areas, science and education cultural areas, residential areas, etc. We also proved the superiority of our method compared with the LDA method based on POI semantic information and the Word2Vec method considering the linear spatial relationship of POI.

Key words: urban functional area identification, contextual relationship, point of interest, spatial co-location mode, Beijing

中图分类号: