The Impact of Visual Language Strategies in Public Art Creation on Community Spatial Perception and Public Behavior

Artificial intelligence and intelligent algorithmic analysis have become important technical tools for interpreting visual information and modeling human–environment interactions in urban public spaces. As a typical form of visual intervention, public murals and graffiti reshape spatial visual structures, yet their influence on spatial cognition and public behavior has rarely been examined from a computable modeling perspective. This study aims to investigate how visual interventions affect community spatial cognition and public space behavior through quantifiable visual feature modeling.Based on mural and graffiti cases in multiple public spaces in Suzhou, a computational analysis framework is constructed by integrating visual feature encoding, spatial cognition modeling, and behavioral data analysis. Visual attributes such as thematic clarity, compositional order, and color contrast are encoded as feature vectors, while spatial cognition and public behavior are modeled using regression-based analytical methods. A quasi-experimental design with pre- and post-intervention comparisons is adopted, involving 268 participants across residential, campus-adjacent, and transitional public spaces.Experimental results show that structured visual interventions significantly enhance spatial legibility and environmental identification (p < 0.01), while public space usage frequency and behavioral normativity increase by over 20% after intervention. The findings demonstrate that visual interventions can be effectively interpreted and evaluated using computational modeling approaches, providing technical support for intelligent public space design and visual governance.