Ancient Book Image Restoration Using Generative Adversarial Networks

Abstract

As an important carrier of historical and cultural inheritance, the restoration of ancient books is of great significance to the protection of cultural relics and cultural inheritance. However, the traditional repair methods have some problems, such as low efficiency and insufficient precision. In this paper, a deep learning-based restoration method for ancient books is proposed, which is divided into two steps: structure reconstruction and color correction. The structure reconstruction network (SRN) uses line drawing information to ensure the authenticity and structural stability of large-scale content, and the color correction network (CCN) makes local color adjustments to missing pixels, reducing color bias and edge hopping problems. The experimental results show that this method effectively improves the restoration efficiency and image quality, and provides a new technical support for the protection and inheritance of ancient books.