在MPEG压缩域中进行运动对象提取的现状研究  

Status Research of Moving Object Extraction in the MPEG Compressed Domain

作　　者：李鹏[1] 李燕[1] 鲁晓峰[1] 王学通[1] 王理[1] 

机构地区：[1]西安理工大学计算机科学与工程学院,陕西西安710048

出　　处：《现代电子技术》2008年第16期150-153,共4页Modern Electronics Technique

摘　　要：运动对象的提取是进行视频分析的一个重要部分。传统的提取方法是在像素域中完成,将其用于以压缩形式存储的视频则需要耗费大量的时间先进行压缩码流的解码。为了提高速度,许多科研人员开始分析MPEG码流的特性,研究直接在压缩域中进行运动对象提取的方法。对近年来提出的压缩域中运动对象的提取方法进行分析,为进一步提高算法性能奠定基础。Moving object extraction is one of the important procedures in the video analysis. The traditional extraction methods are carried out in the pixel domain. Due to the heavy work to decompress the videos,it＇s a time-consuming work to analyze the videos stored in compressed form by traditional extraction methods. In order to speed up the extraction procedure, many researchers have begun to study the characters of the MPEG coded stream and proposed some new extraction methods which are implemented in the compressed domain directly. In this paper, some recently developed moving object extraction methods are introduced and compared for further improvement of the extraction algorithms.

关 键 词：MPEG码流 压缩域 运动对象提取 视频分析 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]