非压缩域数字视频中文字的检测与提取.pdfVIP

中国体视学与图像分析　2004 年 第 9 卷 第 4 期 244 CHINESE J OURNAL OF STEREOLOGY AND IMAGE ANALYSIS Vol9 No 4 Sept 2004 文章编号 : 1007 - 1482 (2004) 04 - 0244 - 05 ·综述· 非压缩域数字视频中文字的检测与提取 高平利 , 任金昌 , 赵荣椿 (西北工业大学计算机学院 , 西安 710072) 【摘要】: 作为一种高级语义特征 , 视频中的文字信息对视频内容的理解、索引和检索具有重要意 义。本文针对非压缩域中视频文字的检测与提取技术做了详尽的分析和讨论 , 涉及的内容包括 : 文字事件的检测、帧图像中文字区域的检测与分割以及视频文字识别 (Video OCR) 等。对于 不同的典型算法 , 分析了其理论基础和应用特点 , 并且通过对比指出其不足和应用的局限。最 后 , 本文还展望了该技术未来发展的方向。 【关键词】: 非压缩域 ; 文字事件检测 ; 文字分割与识别 ; 视频索引与检索 【中图分类号】: TP39143 　　　　【文献标识码】: A Text Detection and Extraction in Uncompressed Digital Videos Gao Pingli , Ren Jinchang ,Zhao Rongchun ( School of Computer Science , Northwestern Polytechnical University , Xi ’ an 710072) 【ABSTRACT】: As a high - level semantic feature , text information of videos is very important for un derstanding , indexing and retrieval of video contents. In this paper , comprehensive analysis and dis cussions are given on detection and extraction of texts from videos in uncompressed domain , in which several topics are involved : text event detection , text region detection and segmentation in video frames as well as video OCR , etc. As for different typical algorithms , their theoretical basis and ap plications are discussed , along with shortcomings and limitations in applications are pointed out through comparisons. Finally , some relevant prospects are also proposed. 【KEYWORDS】:uncompressed domain ;text event detection ; text segmentation and recognition ; video indexing and retrieval 1 　引言