融合出租车轨迹与街景图像的城市街道空间分类方法  

Urban Function Recognition at Street Level by Integrating Taxi Trajectory and Street-Level Imagery

作　　者：郭海京 钟远军 邢汉发 高绵新 彭嘉茵 Guo Haijing;Zhong Yuanjun;Xing Hanfa;Gao Mianxin;Peng Jiayin(Surveying and Mapping Institute Lands and Resource Department of Guangdong Province,Guangzhou 510500,China;Beidou Research Institute,Faculty of Engineering,South China Normal University,Foshan 528225,China;School of Geography,South China Normal University,Guangzhou 510631,China;Key Laboratory of South China Tropical and Subtropical Natural Resources Monitoring,Ministry of Natural Resources,Guangzhou 510500,China;Natural Resources Science and Technology Collaborative Innovation Center of Guangdong Province,Guangzhou 510500,China)

机构地区：[1]广东省国土资源测绘院,广州510500 [2]华南师范大学北斗研究院,广东佛山528225 [3]华南师范大学地理科学学院,广州510631 [4]自然资源部华南热带亚热带自然资源监测重点实验室,广州510500 [5]广东省自然资源科技协同创新中心,广州510500

出　　处：《热带地理》2024年第5期906-920,共15页Tropical Geography

基　　金：广东省科技计划项目(2021B1111610001、2021B1212100003)。

摘　　要：城市街道空间是一种复杂的公共活动场所,是城市功能的重要空间载体。然而现有研究通常只聚焦于街道空间的交通功能,忽略街道空间所承担的其他功能类型,给街道的设计以及品质优化带来阻碍。为此,文章提出了一种融合出租车轨迹与街景图像的街道空间城市功能分类方法,该方法基于出租车轨迹数据构建城市居民在街道空间上的动态出行特征,基于街景图像构建街道空间的物理环境特征,并采用K-Means聚类算法对街道空间进行聚类。并以深圳市宝安区为例进行实验,结果表明,该方法能识别出商业、交通和居住3种类型的街道空间,识别结果可为街道空间设计与品质优化提供参考。Urbanization in China has entered a new phase that emphasizes both scale expansion and quality improvement.This has led to demands on urban functional structures and rational urban planning.Street space serves as a vital spatial carrier for meeting urban residents'needs,such as travel,shopping,and leisure.It is comprised of urban roads and their ancillary facilities,buildings along the route,and many other elements.However,existing studies typically focus on the traffic function of urban roads,overlooking other functional aspects of street space as complex public activity areas,thereby hindering the optimization of street space quality.Therefore,there is a need to propose a classification method for the urban street space functions.Given the proliferation of taxi trajectory data and street view imagery,street space can be described in detail from the citizens'perspectives.Therefore,this study proposes a street space classification method that integrates taxi trajectory and street view imagery to delineate urban functions.The dynamic travel characteristics of urban residents in the street space are constructed using the taxi trajectory,including the number of trajectories passing by the street,the number of origin points on the street,and the number of destination points on the street.The physical environment characteristics of the street are constructed using the street view image,which contains single element street features,combined element street features,and overall element street features.Subsequently,based on residents'dynamic travel characteristics and the physical environmental characteristics of street space,the K-Means method is utilized to allocate street spaces with similar urban functions into the same clusters.Taking Bao'an District of Shenzhen as a case study,it was found that clustering with K=3 yielded the most interpretable results.Subsequently,based on street characteristics and auxiliary POI information(including POI density and POI enrichment index),street spaces were successfully classified

关 键 词：城市功能 街道空间 出租车轨迹 街景图像 深圳市 

分 类 号：TU984.113[建筑科学—城市规划与设计]