How the characteristics of street color affect visitor emotional experience

Abstract

Streets play a crucial role in urban tourism. This study examines the impact of urban street landscape color characteristics on visitors’ emotional perceptions, providing insights for landscape planning. Focusing on Xi'an's inner-third ring road area, we use Full Convolutional Neural Network (FCN) and Random Forest (RF) algorithms to create an visitors’ emotional perception dataset of street images. Machine learning techniques are employed to extract color features, construct quantitative color indexes, and visualize them spatially. Key findings include: (1) A spatial pattern where perceptions of beautiful and lively increase and depressing decreases from city centers, suggesting similarities between visitors' and residents' emotional responses; (2) A complex, non-linear relationship between color features and emotional perception, with optimal responses at color complexity 0.86 and coordination 0.84; (3) More pronounced color characteristics positively affect visitor emotions under non-routine conditions. Theoretically, this study confirms that vibrant environments enhance visitor experiences. Methodologically, it extends tourism studies by integrating streetscape big data and machine learning, moving beyond traditional text-based analysis. The results offer city managers valuable insights into visual preferences for streetscapes, aiding in the optimization of urban landscape design.