基于深度学习和多源数据的街道美感评价与影响因素分析——以上海为例

Evaluation of Street Aesthetics and Influencing Factors Using Multi-source Deep Learning Methods:The Case Study of Shanghai

营造兼具交通功能和步行空间感知的美学街道空间,对于实现交通发展模式的转变至关重要。地图兴趣点、街景图像、三维建筑地图等数据的出现,为街道空间研究提供了大量新图像和数据支持,打破了数据源的限制。本文以上海市街道空间为研究对象,基于深度学习技术和GIS技术,以多源数据为载体,大规模进行街景美感评估以及街道物理空间、底层界面、绿化设施、宏观形态四类客体指标测度。在此基础上,通过数理推导,揭示客体指标对街道美感评价的关系。研究发现,物理空间对个体在街道三维关系的感知方面影响最大,环境设施和底层界面指标对美感感知影响次之,绿视率与美感是影响美感感知的关键因子,街道宽度、高宽比对美感的感知影响更大,店面招牌个数、建筑密度对于美感评价也具有积极影响。本研究可为优化街道空间布局提供依据,为更好地营造可步行城市与健康城市提供科学理性支撑,为相关学科优化建设环境提供实践支持。

It is crucial to create an aesthetic street space that combines traffic function and pedestrian space perception through the environment,and supports social behavior to realize the transformation of the transportation development model.However,street aesthetic assessment and analysis are limited due to insufficient technology and availability of data sources.The emergence of data such as map points of interest(POI),street view images,and three-dimensional building maps has provided a large number of new images and data support for the street space research,breaking the limitations of data sources.This paper takes Shanghai street space as the research object,based on deep learning and GIS,using multi-source data as the carrier to conduct large-scale streetscape aesthetic evaluation and measurement of four types of object indicators:street physical space,ground floor interface,greening facility,and the macroscopic morphology.On this basis,the relationship between the object indicators and the street aesthetic assessment is revealed through mathematical derivation.It finds that physical space has the greatest influence on the perception of individuals in the three-dimensional relationship of streets;the green viewing rate is a key factor affecting the perception of street aesthetics;the street width has a greater impact on the aesthetic perception than the aspect ratio;the number of storefront signboards and building density also have a positive impact on the aesthetic evaluation.This paper can provide a basis for optimizing the spatial layout of streets,scientific and rational support for creating walkable cities and healthy cities,and practical support for optimizing the built environment in related disciplines.