首页 | 本学科首页   官方微博 | 高级检索  
     检索      

基于Mean-Shift和粒子滤波的两步多目标跟踪方法
引用本文:李红波,曾德龙,吴渝.基于Mean-Shift和粒子滤波的两步多目标跟踪方法[J].重庆邮电大学学报(自然科学版),2010,22(1):112-117.
作者姓名:李红波  曾德龙  吴渝
作者单位:重庆邮电大学人工智能研究所,重庆,400065;重庆邮电大学人工智能研究所,重庆,400065;重庆邮电大学人工智能研究所,重庆,400065
基金项目:重庆市科技攻关项目,重庆市自然科学基金,重庆市教委科学技术项目 
摘    要:针对Mean-Shift跟踪算法容易跟踪丢失以及粒子滤波跟踪算法计算量大等问题,提出了一种两步多目标跟踪算法.利用Mean-Shift进行第一步跟踪得到候选目标,在跟踪不准的情况下再利用粒子滤波进行后续的跟踪结果修正.实验结果表明两步跟踪算法既能保持Mean-Shift跟踪的实时性,也能维持粒子滤波跟踪算法的鲁棒性,有很高的实用性.

关 键 词:Mean-Shift  粒子滤波  两步跟踪  目标跟踪
收稿时间:3/5/2009 12:00:00 AM

A two-step multiple targets tracking algorithm based on Mean-Shift and particle filter
LI Hong-bo,ZENG De-long,WU Yu.A two-step multiple targets tracking algorithm based on Mean-Shift and particle filter[J].Journal of Chongqing University of Posts and Telecommunications,2010,22(1):112-117.
Authors:LI Hong-bo  ZENG De-long  WU Yu
Institution:Institute of Artificial Intelligence, Chongqing University of Posts and Telecommunications, Chongqing 400065, P. R. China
Abstract:Since Mean-Shift (MS) tracking algorithm always loses tracking objects and the particle filter tracking algorithm costs huge computation, a novel two-step multiple targets tracking algorithm was proposed. Firstly, a candidate object was gotten by Mean-Shift (MS) tracking algorithm. Then, the tracking result would be verified by particle filter technique when the object couldn't be traced exactly. Experimental results show that this approach can maintain the efficiency of MS algorithm and the powerful ability of particle filter technique, so it is of high practicability.
Keywords:Mean-Shift
本文献已被 万方数据 等数据库收录!
点击此处可从《重庆邮电大学学报(自然科学版)》浏览原始摘要信息
点击此处可从《重庆邮电大学学报(自然科学版)》下载免费的PDF全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号