作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2011, Vol. 37 ›› Issue (20): 169-171. doi: 10.3969/j.issn.1000-3428.2011.20.058

• 人工智能及识别技术 • 上一篇    下一篇

针对融合图像的均值位移跟踪方法

薛 雷,肖 刚   

  1. (上海交通大学航空航天学院,上海 200240)
  • 收稿日期:2011-04-15 出版日期:2011-10-20 发布日期:2011-10-20
  • 作者简介:薛 雷(1985-),男,硕士研究生,主研方向:图像处理,融合识别;肖 刚,副教授、博士
  • 基金资助:
    国家自然科学基金资助项目(60904096);航空科学基金 资助项目(20095557010, 20102057006)

Mean Shift Tracking Method for Fusion Image

XUE Lei, XIAO Gang   

  1. (School of Aeronautics and Astronautics, Shanghai Jiaotong University, Shanghai 200240, China)
  • Received:2011-04-15 Online:2011-10-20 Published:2011-10-20

摘要: 提出一种针对融合图像的均值位移跟踪方法。运用加权平均融合和基于动态目标检测的多源动态图像融合体系,对可见光与红外图像进行像素级动态图像融合,利用Mean Shift算法进行目标跟踪。采用真实数据进行实验,并对实验结果进行稳态的位置均方根误差评价。结果表明,在跨背景区域及复杂背景下,采用该方法对目标进行运动跟踪,能满足鲁棒性及实时性要求。

关键词: 融合跟踪, 像素级图像融合, 动态图像融合, 目标跟踪, 均值位移

Abstract: This paper proposes a Mean Shift tracking method for fusion image. It uses weighted average fusion and fusion scheme based on region target detection to make pixel-level dynamic image fusion of video from visible and infrared cameras. The tracking method uses Mean Shift algorithm. Fusion experiments results with real world image sequences are evaluated by the Root Mean Square(RMS) error of steady-state. It indicates that aiming at the target moving from one area to a different area, the proposed method is effective and efficient, meeting the robustness and real-time requirements.

Key words: fusion tracking, pixel-level image fusion, dynamic image fusion, target tracking, Mean Shift

中图分类号: