Research on precise restoration of Thangka and Tibetan murals integrating contour drawing techniques

Simple item page
dc.contributor.authorLi, Mingxuen_US
dc.contributor.editorCampana, Stefanoen_US
dc.contributor.editorFerdani, Danieleen_US
dc.contributor.editorGraf, Holgeren_US
dc.contributor.editorGuidi, Gabrieleen_US
dc.contributor.editorHegarty, Zackaryen_US
dc.contributor.editorPescarin, Sofiaen_US
dc.contributor.editorRemondino, Fabioen_US
dc.date.accessioned2025-09-05T19:06:16Z
dc.date.available2025-09-05T19:06:16Z
dc.date.issued2025
dc.description.abstractChinese murals and Thangka paintings are crucial elements of traditional culture and have occupied a central position in religious practices. However, these artworks are increasingly threatened by natural deterioration and human interventions. Although current restoration methods have adopted data-driven regeneration paradigms, these models are trained to reconstruct appearances based on learned statistical patterns, yet often overlook the symbolic and semantic layers intrinsic to the artworks. In response, we propose a line-drawing guided restoration framework. Functioning as an aid rather than a substitute for human expertise, our model generates structure-preserving suggestions that assist human restorers in accurately reconstructing damaged artworks. The model employs encoders capable of capturing features from both original Thangka images and their corresponding line drawings, mapping them into a discrete latent space for further processing. Subsequently, a decoder synthesizes restored images from these fused representations, achieving faithful inpainting. Evaluation on datasets of Thangka paintings demonstrates that, compared to state-of-the-art methods, our approach achieves maximum reductions of 58.5% in MAE and 63.8% in LPIPS, coupled with SSIM improvement reaching 44.0%. These findings substantiate the efficacy of our method in preserving intricate details and improving visual coherence, supporting more faithful cultural preservation.en_US
dc.description.sectionheadersInstrumental and Computational Approaches for CH Conservation and Restauration
dc.description.seriesinformationDigital Heritage
dc.identifier.doi10.2312/dh.20253242
dc.identifier.isbn978-3-03868-277-6
dc.identifier.pages9 pages
dc.identifier.urihttps://doi.org/10.2312/dh.20253242
dc.identifier.urihttps://diglib.eg.org/handle/10.2312/dh20253242
dc.publisherThe Eurographics Associationen_US
dc.rightsAttribution 4.0 International License
dc.rights.urihttps://creativecommons.org/licenses/by/4.0/
dc.subjectCCS Concepts: Applied computing → Arts and humanities; Computing methodologies → Image processing; Machine learning
dc.subjectApplied computing → Arts and humanities
dc.subjectComputing methodologies → Image processing
dc.subjectMachine learning
dc.titleResearch on precise restoration of Thangka and Tibetan murals integrating contour drawing techniquesen_US
Files
Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
dh20253242.pdf
Size:
17.89 MB
Format:
Adobe Portable Document Format
Download
Collections
Track 01 – Documentation, Preservation, Monitoring and Restoration