城市街道绿化泛类结构的视觉感知特征——以天津市市内六区为例  

Research on Pedestrians’Visual Perception Characteristics Based on USGGS:A Case Study of the Six Inner-City Districts of Tianjin

作　　者：杨瀚文 邹尚恩 胡一可[1,2,3] YANG Hanwen;TSOU Shang’en;HU Yike(the School of Architecture,Tianjin University;the Human Settlements Committee,Chinese Society for Urban Studies;Department of Landscape Architecture,School of Architecture,Tianjin University)

机构地区：[1]天津大学建筑学院 [2]中国城市科学研究会城市人居环境专业委员会 [3]天津大学建筑学院风景园林系

出　　处：《风景园林》2025年第5期37-44,共8页Landscape Architecture

基　　金：国家自然科学基金重点项目“基于中华语境“建筑-人-环境”融贯机制的当代营建体系重构研究”(编号52038007)。

摘　　要：【目的】绿化资源配置是城市公共空间优化的重要环节之一,对居民生活质量的提升有着积极的作用。城市街道绿化泛类结构(urban street greening general structure,USGGS)能够反映街道绿化在行人视觉环境中的整体特征,研究USGGS聚类对于物质空间要素数量以及物质空间形态的改变,能够有效探究街道绿化对行人视觉感知水平的影响。【方法】采用百度街景数据,利用DeepLabV3+神经网络模型,对天津市市内六区街道的物质空间要素进行分割,使用ArcGIS软件对空间分布特征进行可视化处理,结合数理统计分析结果,探讨USGGS与行人视觉感知之间的关系。【结果】USGGS聚类呈现向心聚集型的空间分布特征,城市主干道及快速路的行人视觉感知空间分布特征较为同质化,空间异质化现象集中出现在街道断面狭窄的生活型街道以及商业型街道。不同聚类的USGGS不仅对行人视觉感知有不同程度的影响,也与场所属性以及绿化空间位置密切相关。【结论】提升城市街道环境质量需要考虑行人视觉感知水平。合理的USGGS配置以及适当的种植点位能够更好地适应周围场所的属性,促进城市公共空间与城市街道绿化的有机融合,助推城市更新工作的精细化管理,提升城市人居环境质量。[Objective]This research aims to explore in depth the interactions between urban street greening general structure(USGGS),and pedestrians’visual perception,and the influence of USGGS on urban ecological environment and residents’health.USGGS,the vertical hierarchical structure of urban street greening,covers a variety of dimensions such as trees,shrubs,and grasses,which is of great importance for enhancing the quality of urban ecological environment,improving the physiological health of residents,and alleviating the tension between urban residents and the natural environment.The core objective of the research is to propose strategies to optimize the greening structure of urban streets by analyzing the correlation between USGGS and visual perception of pedestrians,so as to enhance the quality of the visual environment of street pedestrians,improve the urban human settlement environment and promote the organic integration of the urban ecological and humanistic environments,thus providing a new way of thinking for urban planning and construction,and promoting the high-quality and sustainable development of urban street space.[Methods]The research selects the six inner-city districts of Tianjin City as the research area,utilizes Baidu street view image(SVI)as data source,and collect SVIs and their coordinates in the six districts in 2019 by calling Baidu API,with a total of 17,326 points selected and 13,281 valid samples obtained after screening.The DeepLabV3+neural network model is used for semantic segmentation of SVIs to accurately recognize and segment various landscape elements in urban SVIs.The model increases the sensory field of the convolution kernel by Atrous convolution technique and null convolution,allowing the model to capture image details at different resolutions and providing more accurate feature recognition support for subsequent SVI segmentation tasks.Based on the study of pedestrians’visual perception,green view index(GVI),openness,and enclosure are selected as quantitative indicators.Throu

关 键 词：风景园林 计算机视觉 城市街道 绿化结构 视觉感知 空间异质性 

分 类 号：TU984[建筑科学—城市规划与设计]