Scholarlink

<p>High-quality landscapes in urban waterfront parks are essential for enhancing residents’ physical and mental well-being. Traditional visual perception-based landscape assessments often rely on subjective evaluations, which are prone to individual biases and thus result in findings that lack universality and comparability. Eye-tracking experiments and semantic segmentation techniques offer quantitative data to address this limitation. This study examines the Hun River Core Waterfront Park in Shenyang, China. It recruited 32 participants and selected 40 photographs of typical scenes, integrating three established evaluation methods—semantic segmentation, eye-tracking experiments, and public preference evaluation—to establish a multidimensional “objective-physiological-subjective” assessment framework. This approach systematically analyzes visual perception characteristics of landscape elements. The results show the following: (1) There is a contradiction between visual attention to landscape elements and public preference: artificial elements received significantly higher visual attention than natural elements, but their public preference scores were generally low, whereas natural elements scored higher in terms of aesthetic quality and fatigue recovery. (2) Eye-tracking data reveal the mediating role of cognitive load. Artificial landscapes triggered longer first fixation durations and larger pupil diameters, exhibiting characteristics of high cognitive load processing, suggesting they require more effortful processing, while natural landscapes showed the opposite pattern. (3) Structural Equation Modeling further confirmed that eye-tracking indexes play a relatively significant mediating role between objective landscape characteristics and public preference, revealing a complete transmission pathway of “environmental stimulus - visual cognitive processing - subjective evaluation. " This study established a multidimensional evaluation framework for temperate urban waterfront parks, which to a certain extent overcomes the limitation of single-dimensional methods that cannot capture both landscape visual features and human perceptual responses. The framework offers practical suggestions for landscape planning and design, such as increasing the proportion of natural elements and improving the visual coordination of artificial features.</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

Evaluation of landscape elements in urban waterfront parks based on visual perception: construction and validation of a multi-dimensional framework integrating eye-tracking experiments and semantic segmentation

  • 期刊论文

  • 发表日期：

    2026年5月11日

查看全文
  • Dewen Wu,
  • Xing Xiao,
  • Lingtong Deng,
  • Wen Wu

摘要

High-quality landscapes in urban waterfront parks are essential for enhancing residents’ physical and mental well-being. Traditional visual perception-based landscape assessments often rely on subjective evaluations, which are prone to individual biases and thus result in findings that lack universality and comparability. Eye-tracking experiments and semantic segmentation techniques offer quantitative data to address this limitation. This study examines the Hun River Core Waterfront Park in Shenyang, China. It recruited 32 participants and selected 40 photographs of typical scenes, integrating three established evaluation methods—semantic segmentation, eye-tracking experiments, and public preference evaluation—to establish a multidimensional “objective-physiological-subjective” assessment framework. This approach systematically analyzes visual perception characteristics of landscape elements. The results show the following: (1) There is a contradiction between visual attention to landscape elements and public preference: artificial elements received significantly higher visual attention than natural elements, but their public preference scores were generally low, whereas natural elements scored higher in terms of aesthetic quality and fatigue recovery. (2) Eye-tracking data reveal the mediating role of cognitive load. Artificial landscapes triggered longer first fixation durations and larger pupil diameters, exhibiting characteristics of high cognitive load processing, suggesting they require more effortful processing, while natural landscapes showed the opposite pattern. (3) Structural Equation Modeling further confirmed that eye-tracking indexes play a relatively significant mediating role between objective landscape characteristics and public preference, revealing a complete transmission pathway of “environmental stimulus - visual cognitive processing - subjective evaluation. " This study established a multidimensional evaluation framework for temperate urban waterfront parks, which to a certain extent overcomes the limitation of single-dimensional methods that cannot capture both landscape visual features and human perceptual responses. The framework offers practical suggestions for landscape planning and design, such as increasing the proportion of natural elements and improving the visual coordination of artificial features.