Abstract
Urban parks are the primary places where urban residents reconnect with natural environments. Numerous studies have shown that natural landscape features benefit human mental health and promote perceived restoration. However, few studies have explored the extent to which natural landscape features in urban parks can mitigate or compensate for the negative effects of artificial landscape features on emotions and restoration. This study employed a field survey method, collecting questionnaire data from 599 participants in four urban parks in Hangzhou, China. The DeepLabV3+ semantic segmentation tool was employed to identify and extract landscape features from panoramic images. Data were statistically analyzed using Structural Equation Modeling (PLS-SEM) to explore the potential moderating effects of different natural landscape features in various environmental contexts on emotions and restoration. The results from the SEM model indicate that the R(2) values for emotional state (ES) and perceived restorative scale (PRS) are 0.359 and 0.643, respectively, demonstrating an acceptable level of explanatory power and fit for the model. The results revealed that not all artificial landscape features negatively affect emotions and restoration. Although Pavement showed a significant negative impact on Perceived Restoration (β = -0.155, p = 0.004), their effect on emotions was not significant (β = 0.111, p = 0.115). Additionally, the study demonstrated that certain natural landscape features, such as the sky and trees, had a significant moderating effect in alleviating the negative emotions caused by artificial landscape features. However, for Perceived restoration, the moderating effect of these natural landscape features was not significant and, in some cases, even exhibited a negative moderating effect. These findings specifically explore how different natural landscape features can offset the adverse effects of artificial landscape features and, to varying degrees, enhance positive emotional responses and improve perceived restoration. The results contribute to understanding the complex dynamics between landscape features, emotions, and restoration in urban parks, offering strategic recommendations for planning, management, and design to create healthier and more restorative urban park environments.