| 摘要: |
| 存量更新时代，城市街道具备满足公众日常疗愈需求的可观潜力，是健康城市中微观层面的高效资源。虽然已有研究将行人作为街道疗愈要素，但缺乏
定量表征，对环境和行人如何共同影响街道疗愈性的机制不明。基于疗愈环境理论，将行人要素纳入核心变量，结合视频语义分割和行人检测技术，对实验图像
中的环境和行人要素进行识别与分析，为探究街道行人数量和活动对疗愈效应的影响提供技术支撑。结果表明，街道人流量对疗愈感知存在影响阈值，小群体行
为和自发性活动与疗愈效应呈显著正相关性，且在街道环境的基础影响效应之上，行人要素因对疗愈感受解释度较高而不容忽视，可为健康疗愈导向的街道设计
和更新提供依据。 |
| 关键词: 风景园林 街道疗愈效应 视频深度学习 行人检测 感知密度 疗愈环境 |
| DOI：10.19775/j.cla.2025.04.0087 |
| 投稿时间：2024-08-03修订日期：2024-09-30 |
| 基金项目:国家自然科学基金面上项目(42271436) |
|
| Research on the Mechanism of Pedestrian Healing Effects on Streets Based on Video Deep Learning |
| XU Danning,,ZHANG Xia*,,FENG Hengjian,,HAN Litao,,TANG Luliang |
| Abstract: |
| In the context of modern high-density urban spaces, healing
environments have garnered widespread attention due to their positive effects on
the mental and physical well-being of urban dwellers. During the era of existing
building stock renovation, urban streets, characterized by high openness, strong
accessibility, and dense distribution, serve as efficient resources and practical
carriers for promoting public health at the micro level within a healthy city.
These spaces possess considerable potential to meet the public's restorative
needs in daily life. Although previous qualitative studies have already identified
pedestrians as elements contributing to the healing quality of streets, they often
lack a quantitative representation, and the mechanisms through which the
environment and pedestrians collectively interact to influence the healing effects
of streets remain unclear. Hence, the main objective of this study is to explore
the factors influencing street healing effects from both human and environmental
perspectives. It seeks to improve and refine pedestrian measurement indicators,
incorporating video deep learning technology to quantify pedestrian elements and
analyze the correlation between pedestrian flow, activities, and healing effects.
The study aims to clarify the joint influence mechanism of street pedestrians
and the environment on healing experiences. Building upon the theory of
healing environments, this study introduces pedestrian factors as core variables,
alongside environmental factors of the street. Using real-life walking videos for
experimentation, combined with semantic segmentation and pedestrian detection
technologies, this study identifies and analyzes both environmental and pedestrian
factors in experimental images. The research provides technological support for
investigating the impact of pedestrian quantity and activity on healing effects. The
results show that: 1) Pedestrian elements are significantly correlated with street
healing effects. Pedestrian flow on streets has a nonlinear impact on healing
perception, with an observable threshold effect. As pedestrian flow increases,
healing effects show a curve pattern of initially increasing, then decreasing, and
later increasing again. In particular, scenarios with moderate pedestrian flow
exhibit the highest healing effects, followed by those with low pedestrian flow,
while scenarios with high pedestrian flow show the lowest healing effects. Both
the degree of activity clustering and the type of activity can influence healing
perception. Small group behaviors and spontaneous activities show a significant
positive correlation with healing effects, while activities such as parent-child
interactions, resting, photography, and dog walking demonstrate notable healing
properties. When pedestrian elements are coordinated with street environmental
elements, individuals are more likely to experience a positive healing sensation due
to the perceived density being appropriate. 2) on top of the foundational influence
of the street environment, pedestrian elements contribute an incremental 21.1%
to the healing effect, which should not be overlooked. 3) Factors related to natural
and social characteristics, as well as street pedestrian variables, are the primary
contributors to enhancing street healing effects, while street traffic characteristics
have a significant negative impact. This research expands the theoretical
understanding of the healing potential of built environments, demonstrating that
healing effects are not solely derived from the street environment, but are also
influenced by dynamic pedestrians. This recognition of healing effects arising from
the interaction between people and the environment breaks away from previous
assumptions that only considered a single street environment, providing a more
accurate reflection of real-world conditions. These findings may serve as a valuable
reference for decision-making in the design and renewal of health-oriented
healing streets, offer new directions for designing street facilities, improving
residents' health levels, and promoting the development of healthy cities. |
| Key words: landscape architecture street healing effects video deep learning pedestrian detection perceived density healing environment |
