面向历史档案图像的文字修复系统架构设计研究  

Architecture Design of Text Restoration System for Historical Document Images

作　　者：孙凯明 郝明 王刚 吕宜光 Sun Kaiming;Hao Ming;Wang Gang;Lv Yiguang(Intelligent Manufacturing Institute,Heilongjiang Academy of Sciences,Harbin 150090,China)

机构地区：[1]黑龙江省科学院智能制造研究所,哈尔滨150090

出　　处：《黑龙江科学》2024年第23期87-89,共3页Heilongjiang Science

基　　金：黑龙江省科学院重点研发计划项目资助(ZDYF2024ZN02)。

摘　　要：历史档案是珍贵的历史资料,因自然侵蚀和人为因素等对历史档案造成损坏的情况屡见不鲜,这为历史研究者带来了极大的困扰。深入剖析当前国内外在历史档案图像文字修复技术方面的研究进展,设计了一套基于深度学习框架的文字修复系统架构。该系统架构能够有效针对历史档案图像中的文字进行修复,同时提出了一种基于OCR识别的文字修复评价方法,可以为历史档案图像的文字修复提供一条切实可行的技术路径,从而推动历史研究的深入发展。Historical documents are the precious historical materials.But they often suffer from damage caused by natural erosion and human factors,posing significant challenges for historical researchers.The study thoroughly analyzes the current research progress in historical archive image text restoration technology at home and abroad,designs a text restoration system architecture based on a deep learning framework.This architecture can effectively restore the text in historical archive images,and proposes a text restoration evaluation method based on OCR recognition,so as to provide a feasible technical path for the text restoration of historical archive images,and promote the further development of historical research.

关 键 词：历史档案图像 文字修复 系统结构 深度学习 OCR 

分 类 号：G250.7[文化科学—图书馆学] TP391.41[自动化与计算机技术—计算机应用技术]