Classical Chinese poetry contains rich natural imagery whose cultural connotations remain underexplored in digital communication contexts—particularly with respect to how such imagery shapes contemporary audience psychology. This study examines the psychological mechanisms through which natural imagery in the Book of Songs activates place attachment in contemporary audiences, integrating AI (Artificial Intelligence) semantic analysis with empirical survey research. A Bidirectional Encoder Representations from Transformers (BERT)-derived model fine-tuned on Classical Chinese corpora was applied to all 305 poems, achieving high-precision imagery recognition across four categories—flora, fauna, landscape, and climate—with an overall F1 score of 0.913 and a human-machine agreement Kappa of 0.887. Survey data from 526 valid respondents were analyzed using structural equation modeling (SEM) and Bootstrap-based mediation testing. Natural imagery perception significantly predicts psychological response (β = 0.721, p < 0.001). Environmental empathy and place attachment each function as significant mediators, accounting for 55.9% and 39.4% of the total effect, respectively. The chain mediation pathway—perception → environmental empathy → place attachment → psychological response—is statistically significant, explaining 27.0% of the total effect, with total indirect effects reaching 68.2%. These findings demonstrate that traditional cultural symbols activate audience cultural identity through layered psychological mechanisms, offering theoretical grounding and practical guidance for the intelligent digital dissemination of classical literature.