Full-Text Search:
Home|Journal Papers|About CNKI|User Service|FAQ|Contact Us|中文
《Journal of Optoelectronics·Laser》 2018-06
Add to Favorite Get Latest Update

Born-digital image text localization using MSER and local binarization

LIU Mei-hua;FU Cai-ming;LIANG Kai-jian;ZHOU Xi-feng;Hunan Institute of Engineering Engineering Training Center;Hunan Institute of Engineering College of Applied Technology;Hunan Institute of Engineering College of Electrical &Information Engineering;  
Born-digital images provide customers with important sematic information.Locating and recognizing born-digital text can help us manage and search web content.In this paper,a maximally stable extremal regions(MSER)and local binarization based method is proposed for born-digital images text localization.First,the MSER method is used to extract connected components in multi-channels of a borndigital image,and the local binarization method is adopted to refine them,which insures high character detection rate and decreases the number of connected components meanwhile.Then the connected components verification is performed by extracting components low-level features.Finally,text strings construction process is used based on the connected components,and the features between strings and related components are analyzed to find some missing characters to get unbroken text regions.The post processing operations are used to generate final text localization results.The proposed method can receive similar high character detection rate of 92.76% to the conventional MSER method,while get fewer number of components for each image as 232.On the text localization standard ICDAR2013 dataset,recall,precision and Fof the proposed method under the new text localization evaluation are 82.28%,89.35%and 85.67%,respectively.Recall,precision and Fof the proposed method with the traditional text localization evaluation are 87.05%,89.42%and 88.22%,respectively.The experimental results indicate the effectiveness of the proposed text localization method.
【Fund】: 湖南省教育厅科研项目(15C0330);; 湖南省湘潭市联合基金(2017JJ4018);; 湖南工程学院科研资助项目
【CateGory Index】: TP391.41
Download(CAJ format) Download(PDF format)
CAJViewer7.0 supports all the CNKI file formats; AdobeReader only supports the PDF format.
©2006 Tsinghua Tongfang Knowledge Network Technology Co., Ltd.(Beijing)(TTKN) All rights reserved