东北大学学报（自然科学版）

东北大学学报:自然科学版 ›› 2013, Vol. 34 ›› Issue (1): 4-8.DOI: -

• 论著 • 上一篇    下一篇

基于随机蕨丛的长期目标跟踪算法

佟国峰，蒋昭炎，谷久宏，庞晓磊   

  1. (东北大学信息科学与工程学院，辽宁沈阳110819)
  • 收稿日期:2012-07-30 修回日期:2012-07-30 出版日期:2013-01-15 发布日期:2013-01-26
  • 通讯作者: 佟国峰
  • 作者简介:佟国峰(1973-)，男，辽宁沈阳人，东北大学副教授，博士.
  • 基金资助:
    国家自然科学基金资助项目(61175031)；教育部基础科研基金资助项目(N110204004).

Long Term Object Tracking Algorithm Based on Random Ferns

TONG Guo-feng， JIANG Zhao-yan， GU Jiu-hong， PANG Xiao-lei   

  1. School of Information Science ＆ Engineering， Northeastern University， Shenyang 110819， China.
  • Received:2012-07-30 Revised:2012-07-30 Online:2013-01-15 Published:2013-01-26
  • Contact: TONG Guo-feng
  • About author:-
  • Supported by:
    -

RichHTML

1

PDF (PC)

1359

摘要： 为了从根本上解决运动目标遮挡、环境光照变化、目标外观变化、目标运动速度过快等复杂情况下的目标跟踪问题，采用了将目标检测与标准的目标跟踪算法相结合的目标跟踪框架，提出了基于双向一致性误差评估的标准跟踪算法，提高了跟踪点的可靠性；采用随机蕨丛算法作为目标检测的主体，很好地解决了目标遮挡、消失等情况导致跟踪失败后无法重新初始化的问题.通过实验验证了所提出的目标跟踪算法能实现目标的长期跟踪，且具有很强的适用性.

关键词: 随机蕨丛, 目标检测, 光流法, 目标跟踪, 误差评估

Abstract: In order to realize the object tracking in complex situation such as occlusions， partial and full changing illuminations， appearance change or fast motion， a novel object tracking framework was proposed by adopting the method that combined standard object tracking with object detection. The tracking algorithm based on the error of the evaluation standard two-way consistency was put forward to improve the reliability of the tracking point. The target detection algorithm based on random fern plexus was able to initialize objective to again track object， when the object was sheltered and disappear， failure to track. The proposed object tracking algorithm was proved to be able to achieve the goal of long-term tracking and be very robust.

Key words: random ferns, object detection, optical flow method, object tracking, error evaluation

中图分类号: