Understanding the nonlinear effects of the street canyon characteristics on human perceptions with street view images

Abstract

Human perceptions represent the psychological experiences and feelings of individuals toward the surrounding environment. It is influenced by various physical elements (e.g., sky, tree, building, etc.) within the street canyon. However, prior research has predominantly relied on field surveys, high-cost methods, and restricted data sources, thereby limiting analyses of the impact of street canyon characteristics on human perceptions. Importantly, the nonlinear effects of street canyon characteristics on human perceptions have not been fully understood by existing studies. Thus, this study employed Baidu Street View images to evaluate street canyon characteristics and human perceptions at a visual level. A random forest regression was then utilized to uncover the nonlinear effects of street canyon characteristics on human perceptions. The results suggested that some street canyon characteristics nonlinearly affected human perceptions. Specifically, the optimal sky proportion within the individual visual field was 15%, and the appropriate thresholds for trees and grass were 17.5% and 1.0%, respectively. Overall, positive perceptions might be increased if the proportion of natural characteristics within street canyons was maintained below the optimal threshold. In addition, threshold effects have been observed in relation to the perception of beauty due to cars, the perception of vitality and security due to roads, and the perception of boredom due to walls. The findings of this study can serve as scientific evidence and inform urban renewal based on a people-oriented approach.

Keywords

• Human perceptions
• Baidu Street View
• Street environment
• Random forest
• Nonlinear effect