月 计 算 机 学 报 CHINESEJOURNAL OF COMPUTERS Vol 27 No.pdfVIP

第 27 卷 　第 8 期 计 　　算 　　机 　　学 　　报 Vol. 27 No. 8 2004 年 8 月 CHIN ESE JOURNAL OF COMPU TERS Aug. 2004 视频镜头时域分割方法的研究 朱　曦　　林行刚 (清华大学电子工程系图像图形研究所　北京 100084) 摘　要 　视频时域分割指将视频序列分成若干镜头 ,是视频内容分析以及基于内容的视频浏览和检索的第一步. 该文首先对视频结构以及视频镜头种类进行了简要的描述 ,然后对为计算不连续性而采用的提取特征和建立测量 准则的常用方法进行概述. 其后 ,文章介绍了检测镜头切变和渐变的算法及其优缺点. 在压缩域上检测镜头变换边 界的问题也在文中予以分析. 在结论与展望中 ,提出了一些这一领域的难点和对今后工作的展望. 关键词 　视频时域分割 ;切变 ;渐变 ;特征提取 ;压缩域 中图法分类号 TP391 Survey on Video Temporal Segmentation ZHU Xi 　L IN XingGang ( ) Instit ute of I mage and Graphics , Depart ment of Elect ronic Engi neeri ng , Tsi nghua U niversity , Beiji n g 100084 Abstract 　Video temporal segmentation denotes partitioning a video sequence into shots which is the first step toward video content analysis and contentbased video browsing and retrieval. A video shot is defined as a series of interrelated consecutive frames taken contiguously by a single camera. Video shots are also considered to be the primitives for higher level content analysis , indexing , and classifica tion. Starting from a brief description of the structure of video and the classification of video shots , the paper generalize the common methods to extract features and build metrics for computing the disconti nuit y values. The characteristics of the features are presented in details and compared with each other. Afterwards the paper reviews the algorithms for detecting the cut and gradual shot changes and their advantages and d