Digital restoration and feature recognition of a Qing-Dynasty vernacular dwelling based on multimodal data fusion

Abstract

Abstract This study addresses the urgent challenge of digitally preserving severely damaged vernacular architecture that lacks complete historical documentation. Taking the Dan Tao’s Former Residence, a Qing Dynasty dwelling in the Jingchu region, as a case study, we propose a reproducible multimodal framework for reverse restoration. The approach integrates SLAM-based laser scanning, UAV photogrammetry, historical documentary evidence, analogy-driven HBIM construction, and knowledge-graph visualization. By bridging the semantic gaps between material remains, images, and textual records, the workflow enables high-fidelity digital modeling of complex components while decoding cultural features at multiple scales. Results demonstrate that the framework overcomes limitations of insufficient point-cloud density and missing documentation, achieving accurate 3D restoration of degraded structures and establishing a scalable cultural feature recognition system for Jingchu vernacular architecture. This research provides both methodological innovation and practical tools for conservation, offering a transferable paradigm for safeguarding endangered architectural heritage worldwide.