ScholarWorks@중앙대학교

초록

This study integrates Cesare Brandi's restoration theory, UNESCO photograph conservation guidelines, and ICOM-CC international definitions to establish three essential conditions for photograph restoration: scientific foundation, transparency and explainability, and critical expert intervention. Based on these conditions, the research proposes the STC AI Restoration model, a new paradigm that distinguishes itself from existing AI approaches. Through case analysis of Korean War veteran and independence activist photograph restorations, this study demonstrates that current AI technology compromises restoration authenticity by disregarding the physical and chemical deterioration mechanisms of photographs and relying solely on statistical pattern recognition. Expert interviews from photograph conservation, AI technology, and digital archiving fields both validate these limitations empirically and elucidate the systematic direction of STC AI Restoration. The research presents five concrete policy recommendations —including national photograph damage database construction, XAI technology development, and interdisciplinary collaboration frameworks—alongside a five-stage implementation process, offering substantive approaches for AI technology to contribute to the scientific preservation of historical records.

키워드