Scholarlink

<p>Environmental color analysis has advanced remarkably, from analyzing general images using traditional methods to analyzing street-view images with innovative technologies, progress that provides new opportunities for understanding color characteristics. Our study investigates rural environments using deep learning DeepLab V3, comparing color consistency between street-view and pedestrian-view images. We conducted <i>K</i>-means clustering with different <i>K</i> values, converted the images into color values. Results reveal street-view images accurately depict overall environmental colors, while pedestrian-view images capture specific object colors. The production landscape images showed a high correlation between street-view and pedestrian-view at <i>K</i> = 50 in terms of hue, at all <i>K</i> values in terms of saturation, and at <i>K</i> = 20, 30, 50, or 70 in terms of value. This research aids in selecting suitable image types for environmental studies, facilitating more informed color planning and suggestions.</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

Comparison of street and pedestrian views: categorizing environmental color using deep learning

  • 期刊论文

  • 发表日期：

    2025年1月7日

查看全文
  • Pei-Yi Weng,
  • Li-Chih Ho,
  • Dongying Li,
  • Yen-Cheng Chiang

摘要

Environmental color analysis has advanced remarkably, from analyzing general images using traditional methods to analyzing street-view images with innovative technologies, progress that provides new opportunities for understanding color characteristics. Our study investigates rural environments using deep learning DeepLab V3, comparing color consistency between street-view and pedestrian-view images. We conducted K-means clustering with different K values, converted the images into color values. Results reveal street-view images accurately depict overall environmental colors, while pedestrian-view images capture specific object colors. The production landscape images showed a high correlation between street-view and pedestrian-view at K = 50 in terms of hue, at all K values in terms of saturation, and at K = 20, 30, 50, or 70 in terms of value. This research aids in selecting suitable image types for environmental studies, facilitating more informed color planning and suggestions.