Urban green spaces, vital public infrastructure, have received limited research on how their morphology affects visual perception preferences. Using data from ten parks, we generated green space maps from high-resolution satellite imagery and calculated indicators, such as quantity, fragmentation, connectivity, and shape complexity. By combining the Mask2Former image segmentation deep learning model with a multi-objective regression model and structural equation modeling, we analyzed the relationship between green space morphology and visual perception preferences, controlling for geographic and demographic factors. The results showed that green spaces with tighter connectivity, aggregation, continuity, and shape complexity led to more distinct visual perceptions. This relationship was mediated by the proportion of landscape elements. The distribution, shape, and connectivity of urban green spaces had an independent impact on individual visual perception, far exceeding the influence of quantity alone. The spatial morphology of urban green spaces should be incorporated into health-oriented urban space design, exploring the global interest in how green spaces impact urban human well-being, and providing valuable insights for urban green space planning and health-driven urban space design.

This study examines how visual features and green space morphology jointly shape restorative perception in dog-friendly urban green spaces using a data-driven analytical framework. A self-constructed dataset integrating street-view imagery, landscape element composition, and morphological metrics was developed to quantify visual entropy, visual richness, and spatial structure. Ten dimensions of visual perception were modeled using an XGBoost framework optimized with a genetic algorithm, achieving high predictive performance (R² = 0.827–0.989). Streetscape analysis revealed relatively stable visual entropy but pronounced heterogeneity in visual richness, reflecting variability in color, form, and spatial layering. Element-level decomposition showed the visual dominance of natural components, particularly trees, sky, and grass. Piecewise linear regression further identified threshold-dependent and dimension-specific effects of green space proportion, fragmentation, patch size, connectivity, aggregation, and shape complexity. Moderate fragmentation and aggregation enhanced perceived complexity and stimulation, whereas excessive shape complexity reduced most restorative responses.

Urban park assessment typically emphasizes vegetation quantity, yet spaces with similar greenness may still produce different visual experiences. This study aimed to identify the structural characteristics of horticultural configurations and examine their incremental predictive value for multidimensional visual perception. A total of 500 local landscape scenes were selected from 10 urban parks in Chengdu and presented as standardized photographs. One hundred and fifty participants rated 17 visual attributes on a seven-point scale. Landscape structure metrics, principal component analysis, cluster analysis, cross-park validation, and SHAP analysis revealed that horticultural configurations varied primarily along three gradients: vegetation continuity–fragmentation, geometric complexity, and patch size. These gradients formed three representative states: dense woody, open heterogeneous, and impervious fragmented configurations. Visual perception was organized into three functional domains: restorative immersion, spatial openness, and affective richness. Compared with models containing only the NDVI and greenspace coverage, adding configuration metrics increased the cross-park predictive R² by 0.257, 0.181, and 0.241 for the three perceptual domains, respectively. Vegetation aggregation and connectivity, mean patch area, and shape complexity were the primary predictors of the respective domains and exhibited stable nonlinear response patterns. This study extends park assessment from vegetation quantity to spatial organization mechanisms, providing an evidence base for perception-oriented horticultural design and helping reduce costly trial-and-error and inefficient investment in park renewal.

Urban greening studies have traditionally emphasized vegetation quantity, while the perceptual implications of horticultural composition and spatial configuration remain less clearly understood. This study investigated how horticultural configurations were associated with multidimensional landscape perception across 10 urban parks in Chengdu, China. A total of 500 standardized landscape images, comprising 50 images per park, were assessed by 150 participants across 17 perceptual dimensions. Vegetation composition, spatial configuration, and conventional greenness indicators were quantified from the images and park-level spatial data. Principal component analysis and PERMANOVA were used to characterize structural gradients among parks, while exploratory factor analysis was used to identify integrated perceptual domains. XGBoost models with five-fold cross-validation were subsequently developed to predict individual perceptual outcomes, and SHapley Additive exPlanations (SHAP) were applied to interpret feature contributions and nonlinear response patterns. Multi-objective optimization was further used to explore hypothetical configuration scenarios involving trade-offs among perceptual objectives. The first two structural principal components explained 65.60% of the variation in horticultural composition and spatial organization, and park configurations differed significantly (PERMANOVA: F = 166.66, p = 0.001). Horticultural typologies differed across 15 of the 17 perceptual dimensions (all p < 0.001), with the largest differentiation observed for nature (ε² = 0.270), species richness (ε² = 0.259), openness (ε² = 0.251), being away (ε² = 0.245), serenity (ε² = 0.236), and enclosure (ε² = 0.234). The XGBoost models showed consistently strong predictive performance, with test-set R² values of 0.692–0.812 and cross-validated R² values of 0.756–0.796. SHAP analyses indicated differentiated and nonlinear associations: vegetation continuity and enclosure were primarily related to restorative immersion, open configurations to spatial accessibility, and ornamental and compositional complexity to affective richness. The optimization results further identified continuous trade-offs among restorative immersion, spatial openness, and affective richness, rather than a single universally optimal configuration. These findings show that landscape perception is associated with the spatial organization and composition of horticultural elements through multidimensional and nonlinear pathways. The study therefore extends vegetation-quantity assessments by providing a configuration-based analytical framework for exploring perceptually balanced urban horticultural environments.