Modern Social Science Research

The large-scale application of generative AI has provided new pathways for the visual transformation of cultural heritage. However, general-purpose large models tend to understand Jingchu cultural imagery only at the level of superficial symbol collage, resulting in varying degrees of semantic drift. Existing research lacks systematic extraction of visual genes and a mapping mechanism from cultural semantics to visual symbols. This study selects six representative Jingchu heritage object categories and introduces the signifier–signified framework of design semiotics to conduct systematic symbol deconstruction across four dimensions of form, color, meaning, and composition, extracting core visual gene sequences classified into three stability grades. On this basis, a three-level semantic mapping framework of "cultural imagery–visual prototype–design symbol" is constructed, defining the ethical boundaries of visual translation through the demarcation of "invariable essence" and "variable application." The study further discusses the application potential of this framework as a semantic constraint foundation in multi-agent collaborative visual translation scenarios, providing a systematic design methodology for the precise visual expression of Jingchu cultural heritage.

Jingchu cultural heritage; Visual genes; Design semiotics; Semantic mapping; Visual translation