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 ›› 2010, Vol. 36 ›› Issue (22): 228-230. doi: 10.3969/j.issn.1000-3428.2010.22.082

• 多媒体技术及应用 • 上一篇    下一篇

基于多特征融合的粒子滤波视频跟踪算法

刘一鸣,周尚波   

  1. (重庆大学计算机学院,重庆 400044)
  • 出版日期:2010-11-20 发布日期:2010-11-18
  • 作者简介:刘一鸣(1986-),男,硕士研究生,主研方向:视频目标跟踪;周尚波,教授、博士生导师
  • 基金资助:

    重庆市计算机网络与通信技术重点实验室基金资助项目(CY-CNCL-2008-02)

Video Tracking Algorithm of Particle Filtering Based on Multi-feature Fusion

LIU Yi-ming, ZHOU Shang-bo   

  1. (College of Computer, Chongqing University, Chongqing 400044, China)
  • Online:2010-11-20 Published:2010-11-18

摘要:

将颜色特征和边缘特征融合在粒子滤波跟踪框架中,并在原有跟踪算法的基础上提出改进算法,加入许多优化机制,包括利用HSV颜色模型对目标颜色特征进行核密度无参估计,使用更符合目标实际运动特性的动态模型以及利用均值偏移聚类粒子等。同时,在边缘特征匹配中引入均值偏移,通过加入边缘预处理过程,使各粒子权值的分布更加符合实际情况。实验结果表明,该算法具有较好的鲁棒性和实时性。

关键词: 视频跟踪, 粒子滤波, 均值偏移, HSV模型, 边缘匹配

Abstract:

This paper fuses the features of both color and edge in particle filter tracking framework, and proposes an improved tracking algorithm by introducing several optimization factors, such as describing the color feature of target using kernel density estimation under HSV color model, using more practical motion model in particle filtering, clustering the particles with mean shift algorithm, and reordering the steps of clustering and resampling. Additionally, it also improves the edge matching method by introducing mean shift. Such method adds a pretreatment of the edge information into the matching process that makes the distribution of particle weights more rational. Experimental results show this algorithm has better robustness and real-time ability.

Key words: video tracking, particle filtering, mean shift, HSV model, edge matching

中图分类号: