计算机工程与应用 ›› 2007, Vol. 43 ›› Issue (25): 21-24.

• 博士论坛 • 上一篇    下一篇

一种多特征融合的粒子滤波跟踪新算法

王 欢,任明武,杨静宇   

  1. 南京理工大学 计算机科学与技术学院,南京 210094
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2007-09-01 发布日期:2007-09-01
  • 通讯作者: 王 欢

New particle filter tracking method based on multi-features fusion

WANG Huan,REN Ming-wu,YANG Jing-yu   

  1. School of Computer Science,Nanjing University of Science & Technology,Nanjing 210094,China
  • Received:1900-01-01 Revised:1900-01-01 Online:2007-09-01 Published:2007-09-01
  • Contact: WANG Huan

摘要: 仅利用单一的目标特征进行跟踪是大多数跟踪算法鲁棒性不高的重要原因。提出了一种有效的多特征融合跟踪方法,该方法同时结合了颜色和运动边缘特征,并通过粒子滤波方法合理地进行概率融合。实验结果表明,算法能够在一种特征受到背景干扰导致目标鉴别能力丧失时,其它特征仍能稳定可靠地跟踪目标,算法简单,鲁棒性高,能够有效适用于复杂背景下的目标跟踪。

关键词: 目标跟踪, 粒子滤波, 多特征融合

Abstract: Object tracking by using single object feature often results in a poor performance in robustness.In this paper,an object tracking method based on multi-features fusion is proposed,it utilizes both color and motion edge feature,and performs a probability fusion in particle filter framework.Experiment results show that by using the proposed method,when one of the features loses discrimination ability for tracking,other features are still able to track the object stably.Moreover,it is simple and very suitable to be applied to deal with tracking problems in complex scene.

Key words: object tracking, particle filter, multi-features fusion