ScholarWorks@중앙대학교

초록

This study proposes an artificial intelligence–based framework for restoring damaged painted cultural heritage to facilitate commercial design generation. Painted cultural heritage often suffers quality degradation in generated designs owing to accumulated damage, such as surface contamination and discoloration. To address this issue, we incorporate a preprocessing stage using inpainting techniques that automatically restore damaged areas using contextual information from surrounding areas before applying generative design algorithms. Various damage types were addressed using open-source inpainting tools, followed by a comparative analysis of designs generated from original damaged and restored paintings. Results indicate that inpainting preprocessing significantly enhances the visual quality of designs. Designs generated from restored paintings were successfully applied across product categories—including textiles, packaging, and interior materials—satisfying contemporary market standards. This methodology offers an approach to reinterpreting cultural heritage in modern commercial design, highlighting the potential for creating viable products while preserving cultural value.

키워드