A SIFT-based approach of recognition of remotely mobile phone captured text images

2015 International Conference on Advanced Technologies for Communications (ATC) Pub Date : 2015-10-01 DOI:10.1109/ATC.2015.7388337

Binh Quang Long Mai, T. Huynh, Anh Dong Doan

{"title":"A SIFT-based approach of recognition of remotely mobile phone captured text images","authors":"Binh Quang Long Mai, T. Huynh, Anh Dong Doan","doi":"10.1109/ATC.2015.7388337","DOIUrl":null,"url":null,"abstract":"Mobile phones have been the trending and indispensable devices to many people these days. With the advancement of science and technology, inexpensive and versatile phones can be easily found at one's disposal. The field of Optical Character Recognition (OCR) is also promoted, as OCR mobile applications are increasingly chosen. This paper addresses an approach to solve one challenging issue of OCR, which is, recognition of text images taken by mobile phones at remote distances. These images usually appear in small font size and at low resolution, as a result, yield very poor outcomes when fed to OCR systems. The remote text images are first analyzed using the vertical projection profile. Then the words and characters are extracted, respectively, after determining the location of peaks and space in the graph. Finally, SIFT algorithm is applied in recognition stage, using the low resolution text templates collected, to find the best match for the texts. The results showed an improved performance in OCR recognition rate, compared to OCR reading software.","PeriodicalId":142783,"journal":{"name":"2015 International Conference on Advanced Technologies for Communications (ATC)","volume":"284 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2015-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"2","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2015 International Conference on Advanced Technologies for Communications (ATC)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/ATC.2015.7388337","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}

引用次数: 2

Abstract

Mobile phones have been the trending and indispensable devices to many people these days. With the advancement of science and technology, inexpensive and versatile phones can be easily found at one's disposal. The field of Optical Character Recognition (OCR) is also promoted, as OCR mobile applications are increasingly chosen. This paper addresses an approach to solve one challenging issue of OCR, which is, recognition of text images taken by mobile phones at remote distances. These images usually appear in small font size and at low resolution, as a result, yield very poor outcomes when fed to OCR systems. The remote text images are first analyzed using the vertical projection profile. Then the words and characters are extracted, respectively, after determining the location of peaks and space in the graph. Finally, SIFT algorithm is applied in recognition stage, using the low resolution text templates collected, to find the best match for the texts. The results showed an improved performance in OCR recognition rate, compared to OCR reading software.

查看原文

微信好友朋友圈 QQ好友复制链接

本刊更多论文

基于sift的远程手机文本图像识别方法

如今，手机已经成为许多人的潮流和不可或缺的设备。随着科技的进步，人们可以很容易地找到便宜而多功能的手机。光学字符识别(OCR)领域也得到了推动，因为OCR移动应用的选择越来越多。本文提出了一种方法来解决OCR的一个具有挑战性的问题，即识别手机在远距离拍摄的文本图像。这些图像通常以小字体和低分辨率显示，因此，当输入OCR系统时产生非常差的结果。首先利用垂直投影剖面对远程文本图像进行分析。然后，在图中确定峰值和空间的位置后，分别提取单词和字符。最后，在识别阶段应用SIFT算法，利用采集到的低分辨率文本模板，寻找文本的最佳匹配。结果表明，与OCR读取软件相比，OCR识别率有所提高。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文去求助

来源期刊

2015 International Conference on Advanced Technologies for Communications (ATC)

自引率

0.00%

发文量