Research on symbol deconstruction and semantic mapping framework of visual genes of Jingchu cultural heritage
荆楚文化遗产视觉基因的符号解构与语义映射框架研究

摘要

生成式AI的规模化应用为文化遗产视觉转化提供了新路径，但通用大模型对荆楚文化意象的理解停留于表象拼贴，存在不同程度的语义漂移。现有研究缺乏对视觉基因的系统提取及从文化语义到视觉符号的映射机制。本文选取荆楚六类代表性遗产对象，引入设计符号学能指—所指框架，从“形、色、意、构”四维展开符号解构，提取核心视觉基因序列并划分三个稳定性等级。在此基础上，构建“文化意象、视觉原型、设计符号”三级语义映射框架，以“不变之体”与“可变之用”界定转译的伦理边界。研究讨论了该框架在多智能体协同转译场景中的应用潜力，为荆楚文化遗产的精准视觉表达提供系统性设计学路径。

Abstract

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.