| 摘要: |
| 随着中国城镇化转型与城市更新的开展，服务城市
大众美学认知与文化精神的城市风貌与景观塑造成为当前的迫
切需求。以上海典型街景为例，结合问卷调查、深度学习与贝
叶斯统计方法，构建具大范围评估应用潜力的公众审美感知模
型，进行公众景观偏好研究。发现公众对于街景的不同场景审
美感知差异较大；景观美学感知概念模型中关键景观要素为土
地、建筑、树、墙；想象度、自然度、对比度和审美干扰度是
影响公众对景观美学认知的重要指标。研究结果可进一步推广
到大规模公众景观认知快速评估，为城市设计提供本土公众偏
好美学理论的信息支持。 |
| 关键词: 风景园林 景观美学 公众景观偏好 城市街
景 深度学习 贝叶斯网络 |
| DOI： |
|
| 基金项目: |
|
| Study on the Public Perception of Shanghai Urban Streetscape and Landscape Aesthetics Based on Deep Learning |
| QIU Yeshan,CHE Shengquan,XIE Changkun,PAN Haozhi |
| Abstract: |
| With the development of urban urbanization and urban
renewal, urban style and landscape shaping serving the urban
public's aesthetic cognition and cultural spirit has become an
urgent demand. Taking examples of streetscapes in Shanghai, a
combination of questionnaire survey, deep learning and Bayesian
statistical methods are used to build a public aesthetic cognition
model that has potential in evaluating the public landscape
preference in a large scale. The findings suggest an obvious
differences in the aesthetic evaluation of various streetscape scenes;
the key landscape elements in the public aesthetic conceptual model
are land, buildings, trees, and walls; imaginability, naturalness,
contrast, and aesthetic interference are significant indicators that
affect the public's perception of landscape aesthetics. The research
results can be further applied to rapid assessment of large-scale
public aesthetic preferences, providing support for the local public
preference aesthetic theory for urban design. |
| Key words: landscape architecture landscape aesthetics public
landscape preference urban streetscape deep learning Bayesian
network |
