一种基于对象跟踪的视频分割算法  

A Video Segmentation Algorithm Based on Object Tracing

作　　者：李兵[1] 须德[1] 王方石[1] 

机构地区：[1]北京交通大学计算机与信息技术学院,北京100044

出　　处：《北京交通大学学报》2005年第5期89-91,110,共4页JOURNAL OF BEIJING JIAOTONG UNIVERSITY

摘　　要：在新一代MPEG-4视频编码标准中,为了支持面向对象编码和实现基于内容的应用,视频分割成为关键技术之一,而半自动的视频分割是常用而且比较精确的一种分割方法.本文提出一种基于对象跟踪的半自动视频分割算法.该算法在用户的参与下,基于图像的时空信息,进行视频对象(VOP)的分割.实验结果表明,该算法能够较精确地连续分割出视频对象.In MPEG-4 video coding standard, the video segmentation plays an important role in objectoriented coding and the application of content-based technology; semi-automatic video segmentation is a common and acute segment method. A semi-auto algorithm of video segmentation based on object tracing is presented in this paper, In the algorithm, video segmentation makes use of the temporal and spacial information, with the participating of users. The experiments show that the algorithm can precisely and successively tract the video object from video sequences.

关 键 词：视频分割 对象跟踪 半自动 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]