少数民族文字文本分析与识别的研究进展  

Survey on text analysis and recognition for multiethnic scripts

作　　者：王维兰[1] 胡金水 魏宏喜[3] 库尔班·吾布力[4] 邵文苑 毕晓君 贺建军 李振江 丁凯 金连文[10] 高良才[11] Wang Weilan;Hu Jinshui;Wei Hongxi;Ubul Kurban;Shao Wenyuan;Bi Xiaojun;He Jianjun;Li zhenjiang;Ding Kai;Jin Lianwen;Gao Liangcai(School of Mathematics and Computer Science,Northwest Minzu University,Lanzhou 730030,China;iFLYTEK Research Co.,Ltd.,Hefei 230001,China;College of Computer Science-College of Software,Inner Mongolia University,Hohhot 010021,China;School of Computer Science and Technology,Xinjiang University,Urumqi 830046,China;School of Sociology and Political Science,Shanghai University,Shanghai 200000,Chna;School of Information Engineering,Minzu University of China,Beijing 100081,China;College of Information and Communication Engineering,Dalian Minzu University,Dalian 116605,China;School of Cyberspace Security,Gansu University of Political Science and Law,Lanzhou 730000,China;INTSIG Information Co.,Ltd.,Shanghai 200000,China;School of Electronic and Information Engineering,South China University of Technology,Guangzhou 510641,China;Wangxuan Computer Institute,Peking University,Beijing 100871,China)

机构地区：[1]西北民族大学数学与计算机科学学院,兰州730030 [2]科大讯飞研究院,合肥230001 [3]内蒙古大学计算机学院,呼和浩特010021 [4]新疆大学计算机科学与技术学院,乌鲁木齐830046 [5]上海大学社会学院,上海200000 [6]中央民族大学信息工程学院,北京100081 [7]大连民族大学信息与通信工程学院,大连116605 [8]甘肃政法大学网络空间安全学院,兰州730000 [9]上海合合信息科技股份有限公司,上海200000 [10]华南理工大学电子与信息学院,广州510641 [11]北京大学王选计算机研究所,北京100871

出　　处：《中国图象图形学报》2024年第6期1685-1713,共29页Journal of Image and Graphics

基　　金：国家自然科学基金项目(62166036,61772430,62266044,62236011);内蒙古自治区科技计划项目(2019GG281)。

摘　　要：对于少数民族古籍的保护与传承,国家予以高度重视,并强调了对这些不可再生文化资源透彻数字化的重要性。随着文档图像分析与识别技术的不断进步,对少数民族文字的文本分析与识别研究受到广泛关注,并取得显著成就,成为人工智能应用研究的一个热点领域。然而,由于少数民族文字种类繁多、应用场景多样及数据集的稀缺性等问题,这一研究领域仍面临诸多挑战。本文旨在总结先前的工作,并为未来的研究提供支持,重点讨论了印刷体文本、联机手写、古籍文档及场景文字识别等任务,概述了国内外在少数民族文种识别领域的发展和最新成果。首先阐明了少数民族文字文本分析与识别的重要性及其价值,介绍了特定少数民族文字及其古籍文档的特征。然后,回顾了这一领域的发展历史和现状,分析并总结了传统方法的代表性成果及其应用;详细讨论了研究重点向深度神经网络模型和深度学习方法的全面转移,这一转变使得各文种的识别性能得到了显著提升。最后,基于相关分析,本文指出了在不同文种文档分析与识别中存在的精度和泛化能力等方面的不足,以及与汉文文本分析与识别的差异;面对少数民族文字文本识别领域的主要困难与挑战,展望了未来的研究趋势和技术发展目标。China’s ethnic scripts differ in their structure types,creation periods,and regions of usage and scope.The his⁃torical documents and various literary materials written,recorded,and printed in ethnic scripts are even more voluminous,which leave an invaluable wealth for exploring the civilization and development history of different ethnic groups.Com⁃pared with mainstream languages,the study of ethnic minority scripts often faces low-resource conditions.In recent years,the protection and inheritance of the intangible cultural heritage of ethnic minorities have attracted increased attention fromthe country,which has great importance and application value for the protection of irreparable diverse cultural resources.By applying traditional image processing,pattern recognition,and machine learning methods,certain results have beenachieved in text recognition and document recognition in Mongolian,Tibetan,Uyghur,Kazakh,Korean,and other majorlanguages.Compared with mainstream languages such as English and Chinese,the research on the character recognition ofminority languages,the analysis of document images,and the development of application systems is relatively laggingbehind.Since the 21st century,the research and application of ethnic script text analysis and recognition have receivedextensive attention and made remarkable progress due to the continuous development and application of technologies in thefield of document image analysis and recognition.They have become the research hotspots in the field of document analysisand recognition and artificial intelligence.However,a large number of problems still need to be solved in the field of minor⁃ity script text and recognition research due to the large number of minority scripts,the wide range of application scenarios,and the scarcity of datasets.This study reviews the development history and recent progress in this field at home and abroadto better summarize previous works and provide support for the subsequent research.It focuses on four subtasks:printedtext rec

关 键 词：少数民族文字 文档分析与识别 印刷体文本识别 手写识别 古籍文档识别 场景文字识别 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]