Background: Research on spatial imagery as perceived by humans is an important frontier for deepening the theoretical understanding of Tourism Destination Image and promoting sustainable urban development. Significance: This study, from the perspective of tourists, explores the correlation mechanism between the cognitive image and affective image of urban space. This is of great significance for enhancing the overall spatial quality of cities, promoting the integration of the man–land relationship, and driving the sustainable development of tourism. Method: In this study, we took Harbin as the case site, collected 89,375 reviews and 23,561 review images of 488 scenic spots on the Mafengwo and Ctrip platforms, and constructed a multimodal dataset. We classified the image scenes with the help of the Places365-CNN model. We then extracted text emotional features by utilizing the SnowNLP deep learning algorithm. We constructed a map of the spatial influence mechanism acting on cognitive image and emotion through MGWR. Results: The experimental results showed that in the level of Pleasure, the five indicators NHS, HPA, RPA, PDS and WRV had significant spatial correlations with urban sentiment. In the level of Arousal, the three indicators PD, MaSD and WRV showed significant spatial characteristics. Conclusions: This study reveals the influence mechanism of urban spatial perception elements on tourists’ emotions. It not only deepens the understanding of the Tourism Destination Image theory, but also provides a practical path based on the optimization of perception scenarios for the improvement of urban space, which has important implications for regional sustainable development.

Integrating big data analytics with sustainable development offers a new avenue for understanding complex human-environment interactions in urban spaces, particularly in urban tourism planning. However, limited research has examined how user-generated content (UGC) can translate tourists’ perceived destination images into actionable spatial strategies for sustainable urban management. Focusing on the built-up area within Harbin’s Third Ring Road, this study used visual UGC posted by tourists on Mafengwo and Ctrip as its primary analytical material, linking 488 tourism attractions, 89,375 reviews, and 23,561 valid images. Places365-CNN scene recognition, aggregation to 500 m × 500 m grids, and ArcGIS-based spatial analysis were applied to identify and map perceived destination images across 1139 spatial units. The results showed that: First, Harbin’s dominant perceived destination images exhibited a multi-type composite structure. History and Architecture accounted for 39.2%, while Infrastructure and Nature accounted for 13.9% and 12.6%, respectively. Winter ice-and-snow scenes made up 35% of the top 20 scenes, indicating that ice and snow were not isolated attraction symbols but a visual medium permeating both the natural and built environments. Second, perceived destination images varied across urban rings, shifting from a historical and cultural core in the First Ring, through a functional transition zone in the Second Ring, to an ice-and-snow periphery in the Third Ring; the five administrative districts also developed complementary image profiles. Third, the dominant images formed a cross-shaped spatial framework along the Songhua River waterfront and the city’s north–south axis, with Sun Island, Central Street, and the Chinese Baroque Historic and Cultural Block serving as the three main perceptual centers. This study demonstrates how integrating big data analytics, computer vision, and spatial analysis can support smart-city solutions, sustainable destination management, and destination marketing. It provides policymakers with a replicable methodological blueprint for evidence-based spatial planning and consumer-centered resource governance, thereby contributing to greener and more resilient urban futures.

Against the backdrop of population aging, community public spaces are important neighborhood settings for older adults’ daily activities, social ties, and mental well-being. However, few studies have integrated multidimensional perceived environmental attributes, parallel physical-activity and social-participation pathways, and recent positive mental well-being within a single framework. A cross-sectional survey involved 842 adults aged 60 years or older from 15 communities in Shenyang, China. Perceived environmental attributes comprised Functional Supportiveness, Perceived Restorativeness, and Social Interaction Support, with Neighborhood Outdoor Physical Activity and Community-Based Social Participation specified as parallel mediators and mental well-being assessed using the WHO-5 Well-Being Index. Analyses included confirmatory factor analysis, structural equation modeling, bootstrap tests, and multigroup analysis. All three environmental attributes were significantly and positively associated with both community activities and mental well-being, with significant indirect effects through both activity pathways. Among the three environmental attributes, Perceived Restorativeness showed the strongest associations with Neighborhood Outdoor Physical Activity and mental well-being, whereas Social Interaction Support showed the strongest association with Community-Based Social Participation. Age-group differences were non-monotonic and path-specific, while structural heterogeneity was more pronounced across community types. These findings support age-friendly public-space renewal integrating support for use, restorative experiences, and opportunities for social interaction across age groups and community contexts.

The integration of geo-spatial databases, advanced modeling, and artificial intelligence provides novel opportunities to investigate regional differentiation with direct implications for urban science and regional development. To avoid homogeneous competition and support regional tourism planning, it is important to examine how tourists perceive different tourism elements across large regional spaces. However, existing studies still lack a systematic workflow for extracting destination-specific perceptual characteristics and linking them to specific tourism locations. This study develops a UGC-based framework to identify tourism elements and perceptual characteristics, construct POI-level perceptual weights, and map the spatial patterns of tourism perception across Northeast China using 196,028 Ctrip reviews from 1491 POIs. The framework integrates jieba segmentation with BERT ranking for keyword extraction, ChatGPT-assisted filtering and manual review for keyword standardization, K-means clustering for perceptual dimension identification, and GIS-based kernel density analysis for spatial visualization. The main findings are as follows. First, tourism perception in Northeast China presented a diversified composite structure. Second, different perceptual dimensions and representative keywords showed differentiated spatial distributions. Third, the spatial structure of tourism perception in Northeast China was summarized as a composite spatial pattern centered on the Five-Dimensional Composite Perception Core Belt, supported by the Natural-Landscape Perception Belt and the Water Recreation Perception Area, and supplemented by multi-level urban nodes and inter-node perceptual linkages. This study provides methodological support and planning implications for understanding regional tourism differentiation and guiding evidence-based destination planning. The construction of geo-spatial databases plays a fundamental role in linking tourists’ perceptions to specific tourism elements and locations, enabling fine-grained and spatially explicit analysis that is difficult to achieve using conventional survey-based data.