Unveiling greenery and visual comfort: Integrating Green View Index and image segmentation in panoramic rural landscapes

Abstract This study explores the intricate relationship between visual comfort (VICO) and greenery distribution – measured by the Green View Index (GVI) – alongside object composition in the rural landscapes of Ciputri Village, Indonesia. Using panoramic imagery and semantic segmentation models, dominant visual elements were identified, categorized, and analyzed for their contributions to landscape perception. The analysis combined linear regression and Random Forest modelling to evaluate the predictors of VICO. While the linear model showed limited explanatory power (R 2 = 0.37), the Random Forest model (R 2 = 0.60) performed substantially better, highlighting the importance of nonlinear relationships. Key findings reveal GVI as a dominant determinant of VICO, underscoring its substantial role in enhancing visual comfort. Specific objects such as trees, mountains, and the sky positively influenced VICO when visually dominant, whereas walls had the opposite effect. Unique patterns were observed for plants and “null” objects, with their contributions varying according to their positional dominance in the visual composition. For instance, plants in primary positions were associated with reduced VICO due to perceived monotony, but they enhanced VICO when secondary or tertiary, reflecting their role as complementary elements that enrich visual diversity. Overall, these results provide a nuanced understanding of the interplay between greenery, object composition, and visual comfort, suggesting that balanced visual diversity and strategic spatial arrangements can moderately shape landscape perception, with direct implications for rural landscape planning in Indonesia.