Scholarlink

<p>While multi-year street view images (SVIs) are increasingly applied to infer human perception of the built environment, it remains challenging to efficiently identify perception changes over time. To this end, this paper introduces a novel end-to-end Perception Changes Assessment Siamese Convolutional Neural Network (PCAS-CNN) model. The model leverages multi-year SVIs and requires a small training sample set, identifying intuitive and temporal variations in human perception of street space. The analysis subsequently explores associations between perception changes and socio-economic factors. A case study in Wuhan, China shows that the proposed PCAS-CNN model achieves a high accuracy and effectively identifies three types of perception changes, i.e., better, unchanged, and worse. The results suggest that Point of Interest (POI) intensity, house prices, urban vitality, and nighttime light intensity changes are significantly associated with perception changes. Notably, the relevance of these factors varies across different regions and points to uneven patterns of urban development. The developed method enables the identification of perception changes with a small annotation dataset, thus offering support for city-wide applications.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Temporal Changes in Human Perception of Street Space: A Street View Images-Based Analysis

  • 期刊论文

  • 发表日期：

    2026年7月21日

查看全文
  • Rui Xiao,
  • Yang Zhou,
  • XingJian Liu,
  • Qiuping Li

摘要

While multi-year street view images (SVIs) are increasingly applied to infer human perception of the built environment, it remains challenging to efficiently identify perception changes over time. To this end, this paper introduces a novel end-to-end Perception Changes Assessment Siamese Convolutional Neural Network (PCAS-CNN) model. The model leverages multi-year SVIs and requires a small training sample set, identifying intuitive and temporal variations in human perception of street space. The analysis subsequently explores associations between perception changes and socio-economic factors. A case study in Wuhan, China shows that the proposed PCAS-CNN model achieves a high accuracy and effectively identifies three types of perception changes, i.e., better, unchanged, and worse. The results suggest that Point of Interest (POI) intensity, house prices, urban vitality, and nighttime light intensity changes are significantly associated with perception changes. Notably, the relevance of these factors varies across different regions and points to uneven patterns of urban development. The developed method enables the identification of perception changes with a small annotation dataset, thus offering support for city-wide applications.