Scholarlink

<p>Jingdezhen export porcelain represents a significant legacy of Sino-Western cultural exchange. However, the inherent fragility of porcelain has resulted in widespread damage to many historical artifacts. In this study, we propose the use of an improved Denoising Diffusion Probabilistic Model (DDPM) to facilitate the intelligent restoration of damaged export porcelain, addressing the inefficiencies and complexities associated with traditional manual restoration techniques. We first compiled a large dataset of classic blue-and-white export porcelain plates, and then employed the improved DDPM to restore their patterns. Furthermore, we developed an expert system capable of rapid and efficient digital restoration. Comparative analyses were conducted between the improved DDPM, the Fast Marching Method (FMM), and Generative Adversarial Networks (GAN). The results demonstrate that the improved DDPM achieves more natural and globally consistent restoration outcomes. This research provides a novel pathway for the digital preservation and transmission of Chinese cultural heritage.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Intelligent restoration expert system design for Jingdezhen export porcelain via improved denoising diffusion probabilistic model

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年6月25日

查看全文
  • Xinhui Kang,
  • Guiyong Yang

摘要

Jingdezhen export porcelain represents a significant legacy of Sino-Western cultural exchange. However, the inherent fragility of porcelain has resulted in widespread damage to many historical artifacts. In this study, we propose the use of an improved Denoising Diffusion Probabilistic Model (DDPM) to facilitate the intelligent restoration of damaged export porcelain, addressing the inefficiencies and complexities associated with traditional manual restoration techniques. We first compiled a large dataset of classic blue-and-white export porcelain plates, and then employed the improved DDPM to restore their patterns. Furthermore, we developed an expert system capable of rapid and efficient digital restoration. Comparative analyses were conducted between the improved DDPM, the Fast Marching Method (FMM), and Generative Adversarial Networks (GAN). The results demonstrate that the improved DDPM achieves more natural and globally consistent restoration outcomes. This research provides a novel pathway for the digital preservation and transmission of Chinese cultural heritage.