基于街景图片的城市相对贫困空间识别与评估--以广州中心城区为例  

IDENTIFICATION AND ASSESSMENT OF URBAN RELATIVE POVERTY SPACE BASED ON STREET VIEW IMAGES:A CASE STUDY OF GUANGZHOU CENTRAL CITY

作　　者：袁媛 沈睿熙[5] 李珊 吴庆瑜 陈曦 YUAN Yuan;SHEN Ruixi;LI Shan;WU Qingyu;CHEN Xi

机构地区：[1]中山大学地理科学与规划学院,广东省城市化与地理环境空间模拟重点实验室 [2]中山大学城市化研究院 [3]中国城市规划学会 [4]中国城市规划学会青年工作委员会 [5]广州开发区规划勘测设计院有限公司 [6]广州大学建筑与城市规划学院 [7]中山大学地理科学与规划学院

出　　处：《城市规划》2025年第4期56-64,共9页City Planning Review

基　　金：国家自然科学基金项目(41871161,52278085,42301182);广东省自然科学基金(2023A1515010704);广东省基础与应用基础研究基金(2022A1515110331)。

摘　　要：随着我国脱贫攻坚的全面胜利,相对贫困问题成为下一步研究的重点。城市内部相对贫困人群集聚的空间,是城市高质量更新和人居环境改善的重点地区。如何有效识别此类空间?传统的社会经济统计、人口普查数据更新周期长,城市多源数据种类和覆盖范围有限,且它们与城市建成环境特征关联度不大,不能为人居环境高质量更新提供研判基础。街道图像数据更新周期短且覆盖尺度小,能及时有效地反映出街道周边地区的社会经济和建成环境特征。本研究以广州市中心城区79个街道813个居委会为案例,尝试辅助利用多源大数据,基于结果导向的卷积神经网络对海量街景图片进行深度学习,构建居委会尺度下的相对贫困空间识别与预测模型,实证表明该方法具备较好的可行性。本研究有助于高质量城市更新地块甄别,有针对性地制定精细化更新空间政策。With the comprehensive victory in China’s poverty alleviation efforts,relative poverty has become the next key research focus.The spatial clustering of relatively poor populations within cities is a critical area for high-quality urban regeneration and improvements in the living environment.How can such spaces be effectively identified?Traditional socio-economic statistics and census data have long update cycles,while multi-source urban data have limitations in both variety and coverage.Moreover,these data are not strongly correlated with the characteristics of the built environment,making them insufficient for assessing high-quality urban regeneration.Street imagery data,on the other hand,have shorter update cycles and finer spatial coverage,enabling timely and effective reflection of the socio-economic and built environment characteristics of surrounding areas.This study takes 79 sub-districts and 813 neighborhood committees in the central urban area of Guangzhou as a case study.Using multi-source big data and a result-oriented convolutional neural network(CNN),we conduct deep learning on massive street-view images to develop a model for identifying and predicting relative poverty spaces at the neighborhood committee level.Empirical results demonstrate the feasibility of this approach.This study contributes to the identification of high-quality urban regeneration areas and supports the formulation of targeted,refined spatial policies for urban regeneration.

关 键 词：相对贫困 街景图片 卷积神经网络 识别与评估 广州中心城区 

分 类 号：TU984[建筑科学—城市规划与设计]