• 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

J4 ›› 2015, Vol. 37 ›› Issue (04): 760-764.

• 论文 • 上一篇    下一篇

视频图像文字检测综述

周东傲,林嘉宇   

  1. (国防科学技术大学电子科学与工程学院,湖南 长沙 410073)
  • 收稿日期:2014-01-03 修回日期:2014-04-23 出版日期:2015-04-25 发布日期:2015-04-25

Video text detection:A review 

ZHOU Dongao,LIN Jiayu   

  1. (College of Electronics Science and Engineering,National University of Defence Technology,Changsha 410073,China)
  • Received:2014-01-03 Revised:2014-04-23 Online:2015-04-25 Published:2015-04-25

摘要:

自动从视频图像中提取文字信息,对于监控视频图像内容、添加视频标签和建立视频图像检索系统,有重要的意义。文字检测是文字信息提取系统的前端,是文字信息提取中最关键的一步。近年来,视频图像文字信息检测领域有了新的重要的发展,综述从基于区域和基于纹理的文字检测方法进行归纳、比较和分析,概括了近年来文字检测技术的主要进展。此外,为了突出综合性方法的重要性,对其专门进行了总结。最后对视频图像中的文字检测技术的难点进行总结,并对其发展趋势进行展望。

关键词: 视频图像文字, 文字检测, 边缘检测, 区域分析, 基于纹理方法, 机器学习

Abstract:

Automatically extracting text information from video images has important significance for video content surveilling,video labeling and video retrieval system.Text detection is the frontend of text information extraction system and the key step of extracting textual information.In recent years,the field of textual information detection in video images has important and new developments.We conduct a review,compare and analyze the regionbased and texturebased methods of text detection.In order to highlight the importance of hybrid approaches,these methods have been specially summarized.Finally,the difficulty of text detection technology in video images has been summarized,and its development trend is discussed.

Key words: video text;text detection;edge-based detection;region-based analysis;texture-based method;machine learning