研究内容

Research contents

老年人分类与步行需求：将老年人根据身体能力分为独立型、辅助型和依赖型三类，并分析他们在步行时对安全、舒适和兴趣的不同需求。

街道建成环境评估：识别并提取街道建成环境的关键因素，如绿化、街道家具、建筑密度等，以客观评估街道步行性。

主观步行感知收集：通过街景图像选择网站，收集三类老年人对不同街景图像的主观感知评分，涵盖安全、舒适和兴趣三个方面。

步行性与建成环境关系：分析客观街道建成环境因素与老年人主观步行感知之间的相关性，探讨如何优化街道环境以提升老年人的步行体验。

Classification and walking needs of the elderly: The elderly are divided into three categories: independent, assisted and dependent according to their physical abilities, and their different needs for safety, comfort and interest when walking are analyzed.

Street built environment assessment: Identify and extract key factors of the street built environment, such as greening, street furniture, building density, etc., to objectively assess street walkability.

Subjective walking perception collection: Through the street view image selection website, the subjective perception scores of three types of elderly people on different street view images are collected, covering three aspects: safety, comfort and interest.

The relationship between walkability and the built environment: Analyze the correlation between objective street built environment factors and the elderly’s subjective walking perception, and explore how to optimize the street environment to improve the walking experience of the elderly.

研究方法

Research methods

分类与样本选择：将老年人按身体能力分类，并通过特定渠道选择样本，确保覆盖不同类型的老年人。

街景图像收集与处理：使用手持设备从行人视角收集街景图像，通过图像分割技术提取街道建成环境元素。

感知数据收集：开发在线图像选择网站，让老年人根据主观感知对街景图像进行评分。

数据分析与模型构建：利用机器学习方法（如梯度提升模型）对收集到的感知数据进行训练，构建预测模型，用于评估所有街景图像的步行性评分。

结果可视化：使用地理信息系统（GIS）软件将预测结果可视化，生成步行性地图，便于直观展示和分析。

Classification and sample selection: Classify the elderly according to their physical abilities and select samples through specific channels to ensure coverage of different types of elderly people.

Street view image collection and processing: Use handheld devices to collect street view images from a pedestrian perspective, and extract street built environment elements through image segmentation technology.

Perceptual data collection: An online image selection website was developed to allow older adults to rate street view images based on subjective perception.

Data analysis and model construction: Use machine learning methods (such as gradient boosting models) to train the collected sensory data and build a prediction model to evaluate the walkability scores of all Street View images.

Visualization of results: Use geographic information system (GIS) software to visualize the prediction results and generate walkability maps for easy visual display and analysis.

研究结果

Research conclusions

步行感知差异：三类老年人在步行感知上存在显著差异。独立型老年人更关注步行的兴趣，而辅助型和依赖型老年人则更重视步行的安全和舒适。

建成环境因素影响：

绿化：对所有类型老年人的步行感知均有正面影响，尤其是在安全和舒适方面。

街道家具：对独立型老年人的步行兴趣有显著正面影响，但对辅助型和依赖型老年人的影响较小。

建筑密度：高建筑密度可能降低老年人的步行安全感，尤其是在独立型和辅助型老年人中更为明显。

交通状况：车辆和摩托车的存在对所有类型老年人的步行安全和舒适均有负面影响。

模型预测效果：梯度提升模型在预测老年人步行感知评分方面表现出色，具有较高的准确性和可靠性。

可视化步行性地图：通过GIS可视化生成的步行性地图，为城市规划者和设计师提供了直观的参考工具，有助于识别需要改进的街道区域，并制定针对性的优化策略。

Differences in walking perception: There are significant differences in walking perception among the three types of elderly. Independent elderly people pay more attention to the interest of walking, while assisted and dependent elderly people pay more attention to the safety and comfort of walking.

The impact of built environment factors:

Greenery: Positive impact on perceptions of walking among all types of older adults, especially in terms of safety and comfort.

Street furniture: has a significant positive impact on walking interest among independent older adults, but has a smaller impact on assisted and dependent older adults.

Building density: High building density may reduce the walking safety of the elderly, especially among independent and assisted elderly people.

Traffic conditions: The presence of vehicles and motorcycles has a negative impact on walking safety and comfort for all types of older adults.

Model prediction effect: The gradient boosting model performs well in predicting the walking perception scores of the elderly with high accuracy and reliability.

Visual walkability map: The walkability map generated through GIS visualization provides urban planners and designers with an intuitive reference tool to help identify street areas that need improvement and formulate targeted optimization strategies.