孟加拉文手写文本识别

Joan Andreu Sánchez, U. Pal
{"title":"孟加拉文手写文本识别","authors":"Joan Andreu Sánchez, U. Pal","doi":"10.1109/ICFHR.2016.0105","DOIUrl":null,"url":null,"abstract":"Handwritten text recognition of Bengali is a difficult task because of complex character shapes due to the presence of modified/compound characters as well as zone-wise writing styles of different individuals. Most of the research published so far on Bengali handwriting recognition deals with either isolated character recognition or isolated word recognition, and just a few papers have researched on recognition of continuous handwritten Bengali. In this paper we present a research on continuous handwritten Bengali. We follow a classical line-based recognition approach with a system based on hidden Markov models and n-gram language models. These models are trained with automatic methods from annotated data. We research both on the maximum likelihood approach and the minimum error phone approach for training the optical models. We also research on the use of word-based language models and character-based language models. This last approach allow us to deal with the out-of-vocabulary word problem in the test when the training set is of limited size. From the experiments we obtained encouraging results.","PeriodicalId":194844,"journal":{"name":"2016 15th International Conference on Frontiers in Handwriting Recognition (ICFHR)","volume":"8 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2016-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"3","resultStr":"{\"title\":\"Handwritten Text Recognition for Bengali\",\"authors\":\"Joan Andreu Sánchez, U. Pal\",\"doi\":\"10.1109/ICFHR.2016.0105\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Handwritten text recognition of Bengali is a difficult task because of complex character shapes due to the presence of modified/compound characters as well as zone-wise writing styles of different individuals. Most of the research published so far on Bengali handwriting recognition deals with either isolated character recognition or isolated word recognition, and just a few papers have researched on recognition of continuous handwritten Bengali. In this paper we present a research on continuous handwritten Bengali. We follow a classical line-based recognition approach with a system based on hidden Markov models and n-gram language models. These models are trained with automatic methods from annotated data. We research both on the maximum likelihood approach and the minimum error phone approach for training the optical models. We also research on the use of word-based language models and character-based language models. This last approach allow us to deal with the out-of-vocabulary word problem in the test when the training set is of limited size. From the experiments we obtained encouraging results.\",\"PeriodicalId\":194844,\"journal\":{\"name\":\"2016 15th International Conference on Frontiers in Handwriting Recognition (ICFHR)\",\"volume\":\"8 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2016-10-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"3\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2016 15th International Conference on Frontiers in Handwriting Recognition (ICFHR)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/ICFHR.2016.0105\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2016 15th International Conference on Frontiers in Handwriting Recognition (ICFHR)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/ICFHR.2016.0105","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 3

摘要

孟加拉语的手写文本识别是一项困难的任务,因为由于存在修改/复合字符以及不同个体的分区书写风格,孟加拉语的字符形状复杂。目前发表的关于孟加拉文手写识别的研究大多是孤立的字符识别或孤立的单词识别,而对连续手写体孟加拉文识别的研究很少。本文对连续手写体孟加拉语进行了研究。我们使用基于隐马尔可夫模型和n-gram语言模型的系统遵循经典的基于线的识别方法。这些模型是用带注释数据的自动方法训练的。我们研究了最大似然方法和最小误差电话方法来训练光学模型。我们还研究了基于词的语言模型和基于字符的语言模型的使用。当训练集的大小有限时,最后一种方法允许我们处理测试中的词汇外单词问题。从实验中我们得到了令人鼓舞的结果。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
Handwritten Text Recognition for Bengali
Handwritten text recognition of Bengali is a difficult task because of complex character shapes due to the presence of modified/compound characters as well as zone-wise writing styles of different individuals. Most of the research published so far on Bengali handwriting recognition deals with either isolated character recognition or isolated word recognition, and just a few papers have researched on recognition of continuous handwritten Bengali. In this paper we present a research on continuous handwritten Bengali. We follow a classical line-based recognition approach with a system based on hidden Markov models and n-gram language models. These models are trained with automatic methods from annotated data. We research both on the maximum likelihood approach and the minimum error phone approach for training the optical models. We also research on the use of word-based language models and character-based language models. This last approach allow us to deal with the out-of-vocabulary word problem in the test when the training set is of limited size. From the experiments we obtained encouraging results.
求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
自引率
0.00%
发文量
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:604180095
Book学术官方微信