Answers are grounded in this paper and cite their sources.
Computational study demonstrates high-accuracy ethnic style recognition in murals using digital twins and knowledge graphs, suggesting improved intelligent cultural heritage preservation.
The preservation and interpretation of multi-ethnic mural heritage are challenged by fragile material carriers, heterogeneous cultural semantics, and the lack of unified digital representation. To address these issues, this study proposes a cultural translation framework based on Digital Twins (DT) and Knowledge Graphs (KG), integrating geometric perception and semantic reasoning into a dual-domain alignment architecture. Multi-view three-dimensional point clouds are processed using PointNet++ to achieve semantic segmentation of ethnic styles, while Graph Attention Networks (GAT) and Cross-modal Contrastive Learning (CCL) establish visual-semantic correspondence within a unified embedding space. Dynamic physical parameters from the digital twin environment are further incorporated into a Relational Graph Convolutional Network (R-GCN) to generate structured cultural translation assertions with enhanced robustness. Experimental results demonstrate style recognition accuracies of 88.45%, 86.70%, and 94.35% for Tibetan, Mongolian, and Uyghur murals, respectively, outperforming representative baseline methods in semantic segmentation and cross-modal alignment tasks. The proposed framework enables closed-loop interaction between geometric reconstruction and semantic inference, providing an efficient engineering paradigm for intelligent cultural heritage preservation. Moreover, the integration of digital twin sensing, multimodal information fusion, and wireless data acquisition offers valuable references for electromagnetic sensing systems, antenna-enabled digital monitoring platforms, and next-generation intelligent perception networks in complex heritage environments.
No takes yet. Share an insight, caveat, or question.
Sun et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: