为探明绿道景观的感知和物质特征对宜跑性的影响，以第一视角拍摄福州环南台岛绿道视频并截取 100 张图像样本，运用语义分割计算景观物质特征，使用主观问卷评价景观感知特征和总体宜 跑性，构建“物质特征 - 感知特征 - 宜跑性”贝叶斯网络。结果表明：自然、安全、美感感知特征可直接正向解释绿道宜跑性，吸引感知特征通过影响美感间接影响宜跑性；过多的建筑有碍自然感知，过高的天空可视率会降低环境美感，过多的墙体和围栏有损环境的吸引力，从而降低宜跑性；植被与自然、安全、美感和吸引感知特征显著相关，还对宜跑性有直接解释作用。基于研 究发现，绿道跑步环境设计应着重营造植物景观，提高植被覆盖率以提升自然性与美感，利用植物美化人工设施并适度遮挡建筑，同时通过植物围合安全空间，优化跑步体验。

In order to explore the influence of greenway landscape features on runnability at the levelof perception and material elements, this study captured first-person perspective videosof the greenway around Nantai Island in Fuzhou and intercepted 100 image samples.Semantic segmentation was employed to calculate the material features of the landscape,while a subjective questionnaire was utilized to evaluate perceptual features and overallrunnability. A Bayesian network model integrating 'material features-perceptual featuresrunnability' was subsequently constructed. The results showed that nature, safety, andaesthetics can directly and positively explain the runnability of greenways, whereasattraction indirectly influenced runnability by influencing aesthetics. At the level of materialelements, excessive buildings negatively impacted naturalness perception, high skyvisibility reduced environmental aesthetics, and redundant walls and fences diminishedenvironmental attractiveness, which in turn reduces the runnability. Vegetation not onlysignificantly explains the potency of nature, safety, aesthetics, and attraction, but alsodirectly explains the runnability. Based on the research findings, the design of greenwayrunning environment should focus on plantscape creation, improving the coverage toenhance naturalness and aesthetics, utilizing plants to beautify the artificial structures andmoderately shade the buildings, and enclosing the safe spaces through plants to improvethe running experience.

TU986

福建省社会科学基金基础研究项目“基于计算机视觉的城市绿道宜跑品质评价方法研究”（编号：FJ2021C069）