中国邮电高校学报(英文版) ›› 2022, Vol. 29 ›› Issue (2): 13-23.doi: 10. 19682/ j. cnki. 1005-8885. 2022. 0012

所属专题: 文化计算专题

• Special Topic: Cultural Computing • 上一篇    下一篇

Back in time: digital restoration techniques for the millennium Dunhuang murals

陶艺峰,宋宇程,徐梦秋,张闯,吴铭,白苏乐   

  1. 北京邮电大学
  • 收稿日期:2022-01-29 修回日期:2022-03-15 出版日期:2022-04-26 发布日期:2022-04-26
  • 通讯作者: 张闯 E-mail:zhangchuang@bupt.edu.cn
  • 基金资助:

    This work was supported by the Ministry of Education-China Mobile Communications (MCM20190701)

Back in time: digital restoration techniques for the millennium Dunhuang murals

  • Received:2022-01-29 Revised:2022-03-15 Online:2022-04-26 Published:2022-04-26
  • Contact: Chuang ZHANG E-mail:zhangchuang@bupt.edu.cn
  • Supported by:
    This work was supported by the Ministry of Education-China Mobile Communications (MCM20190701)

摘要:

In the long history of more than 1 500 years, Dunhuang murals suffered from various deteriorations causing irreversible damage such as falling off, fading, and so on. However, the existing Dunhuang mural restoration methods are time-consuming and not feasible to facilitate cultural issemination and permanent inheritance. Inspired by cultural computing using artificial intelligence, gated-convolution-based dehaze net (GD-Net) was proposed for Dunhuang mural refurbishment and comprehensive protection. First, a neural network with gated convolution was applied to restore the falling off areas of the mural to ensure the integrity of the mural content. Second, a dehaze network was applied to enhance image quality to cope with the fading of the mural. Besides, a Dunhuang mural dataset was presented to meet the needs of deep learning approach, containing 1 180 images from the Cave 290 and Cave 112 of the Mogao Grottoes. The  experimental results demonstrate the effectiveness and superiority of GD-Net.

关键词: image inpainting, image enhancement, mural restoration

Abstract:

In the long history of more than 1 500 years, Dunhuang murals suffered from various deteriorations causing irreversible damage such as falling off, fading, and so on. However, the existing Dunhuang mural restoration methods are time-consuming and not feasible to facilitate cultural issemination and permanent inheritance. Inspired by cultural computing using artificial intelligence, gated-convolution-based dehaze net (GD-Net) was proposed for Dunhuang mural refurbishment and comprehensive protection. First, a neural network with gated convolution was applied to restore the falling off areas of the mural to ensure the integrity of the mural content. Second, a dehaze network was applied to enhance image quality to cope with the fading of the mural. Besides, a Dunhuang mural dataset was presented to meet the needs of deep learning approach, containing 1 180 images from the Cave 290 and Cave 112 of the Mogao Grottoes. The  experimental results demonstrate the effectiveness and superiority of GD-Net.

Key words: image inpainting|image enhancement|mural restoration