地球信息科学学报 ›› 2021, Vol. 23 ›› Issue (5): 785-801.doi: 10.12082/dqxxkx.2021.200353

• 地球信息科学理论与方法 • 上一篇    下一篇

基于街景影像的城市道路空间舒适度研究

杨灿灿1,2(), 许芳年1, 江岭1,2,*(), 王瑞璠1, 尹力1, 赵明伟1,2, 张鲜鲜1   

  1. 1.滁州学院 地理信息与旅游学院,滁州 239000
    2.实景地理环境安徽省重点实验室,滁州 239000
  • 收稿日期:2020-07-08 修回日期:2020-09-03 出版日期:2021-05-25 发布日期:2021-07-25
  • 通讯作者: *江 岭(1987— ),男,安徽寿县人,博士,副教授,主要研究方向为地形建模及城市雨洪模拟。 E-mail:jiangling_xs@163.com
  • 作者简介:杨灿灿(1988— ),女,安徽萧县人,硕士,讲师,主要研究方向为城市地形分析、遥感应用。E-mail:yangcancan77@126.com
  • 基金资助:
    安徽高校自然科学研究项目重点项目(KJ2020A0721);安徽高校自然科学研究项目重点项目(KJ2020A0722);安徽省高校优秀人才支持一般项目(gxyq2019093);中国博士后科学基金(2018M642146);安徽省自然科学基金项目(1808085QD103);江苏省博士后科研资助计划项目(2018K144C);国家级大学生创新创业训练项目(201910377054)

Approach to Quantify Spatial Comfort of Urban Roads based on Street View Images

YANG Cancan1,2(), XU Fangnian1, JIANG Ling1,2,*(), WANG Ruifan1, YIN Li1, ZHAO Mingwei1,2, ZHANG Xianxian1   

  1. 1. School of Geographical Information and Tourism, Chuzhou University, Chuzhou 239000, China
    2. AnHui Province Key Laboratory of Physical Geographic Environment, Chuzhou 239000, China
  • Received:2020-07-08 Revised:2020-09-03 Online:2021-05-25 Published:2021-07-25
  • Contact: JIANG Ling
  • Supported by:
    Key Project of Natural Science Research Project of Anhui Universities(KJ2020A0721);Key Project of Natural Science Research Project of Anhui Universities(KJ2020A0722);Anhui Province Universities Outstanding Talented Person Support Project(gxyq2019093);China Postdoctoral Science Foundation(2018M642146);Program of Provincial Natural Science Foundation of Anhui(1808085QD103);Jiangsu Planned Projects for Postdoctoral Research Funds(2018K144C);National Undergraduate Training Program for Innovation and Entrepreneurship(201910377054)

摘要:

城市道路空间舒适度是综合刻画城市道路环境的量化指标之一,可为城市道路规划建设、城市交通优化、城市环境评估等提供参考。本文以街景影像为基础数据,从空间视角阐述了城市道路空间舒适度内涵,构建了城市道路空间舒适度量化指标,建立了以语义分割和等距采样为基础的城市道路空间舒适度测度方法和空间插值方法,最后以南京市局部城区为样区进行了实验,并分析了城市道路空间舒适度特征。实验结果表明:① 语义分割中主要要素的分割精度满足要求,分割总体精度为84.92%, Kappa系数为80.06%;② 综合考虑数据处理时间和舒适度量化表达程度, 200 m采样间距可以综合描述研究区道路空间舒适度状态;③ 本文提出的双向约束插值方法可消除突变点、梯形插值形态、花斑状纹理等异常插值现象,同时可有效保持道路空间舒适度方向特征;④ 本文提出的测度方法可以有效度量城市道路空间舒适度,测度等级精度达80%;⑤ 城市道路空间舒适度等级与道路等级整体上呈现正相关关系,且舒适度在实验区呈现南高北低、建邺区较高、鼓楼及历史老城区低的空间分布状态。

关键词: 城市道路, 空间舒适度, 街景影像, 等距采样, 语义分割, 指标量化, 空间插值, 空间分布

Abstract:

Urban road spatial comfort is an useful indicator to comprehensively quantify urban road environment, which plays an important role in urban-related researches such as construction planning, environment assessment, and traffic optimization. Based on the data of street view images, firstly, the concept of urban road spatial comfort is defined from the perspective of space vision. To represent the concept accurately, five quantitative indexes including road unobstructed degree, road spaciousness, greening degree of vegetation, proportion of sky view, and space openness, and a model to quantify urban road spatial comfort are constructed. Secondly, the urban road spatial comfort measurement method is established. To establish the model, an approach that comprehensively combines the semantic segmentation of street view images, weight assignment based on entropy weight, and analytic hierarchy process and dual-constrained interpolation is designed. Finally, we take the local urban area of Nanjing as the experimental area to assess the performance of the proposed approach. The results show that (1) The segmentation accuracy of the main elements meets the requirements, with an overall segmentation accuracy of 84.92% and a Kappa coefficient of 80.06%; (2) The 200-m sampling distance can comprehensively reflect the road spatial comfort degree in the study area considering the processing time and the quantitative expression; (3) The dual-constrained interpolation method proposed in this paper can eliminate the abnormal interpolation phenomena such as abrupt point, trapezoid shape, and disordered texture, and effectively maintain the directional characteristics of road spatial comfort; (4) The method proposed in this paper can effectively measure the urban road spatial comfort with an accuracy above 80%; (5) There is a positive correlation between the spatial comfort degree and the road grade, that is to say, the higher the road grade, the higher proportion of high-degree spatial comfort in roads. The spatial comfort of urban roads in the experimental area is higher in the south and lower in the north and is higher in Jianye district and lower in Gulou and historical old urban area.

Key words: urban road, spatial comfort, street view image, equidistance sampling, semantic segmentation, index quantification, spatial interpolation, spatial distribution