[1]刘 清,吴志刚,窦 琴,等.粒子滤波的视频目标跟踪算法研究[J].智能系统学报,2009,4(6):538-543.[doi:10.3969/j.issn.1673-4785.2009.06.012]
LIU Qing,WU Zhi-gang,DOU Qin,et al.A particle filtering algorithm for tracking moving objects in videos[J].CAAI Transactions on Intelligent Systems,2009,4(6):538-543.[doi:10.3969/j.issn.1673-4785.2009.06.012]
点击复制
《智能系统学报》[ISSN 1673-4785/CN 23-1538/TP] 卷:
4
期数:
2009年第6期
页码:
538-543
栏目:
学术论文—机器感知与模式识别
出版日期:
2009-12-25
- Title:
-
A particle filtering algorithm for tracking moving objects in videos
- 文章编号:
-
1673-4785(2009)06-0538-06
- 作者:
-
刘 清,吴志刚,窦 琴,熊 燕
-
武汉理工大学 自动化学院,湖北 武汉 430063
- Author(s):
-
LIU Qing, WU Zhi-gang, DOU Qin, XIONG Yan
-
School of Automation, Wuhan University of Technology, Wuhan 430063, China
-
- 关键词:
-
目标跟踪; 粒子滤波; 积分图; 前景分割; 视频
- Keywords:
-
object tracking; particle filter; integral image; foreground segmentation; video
- 分类号:
-
TP18;TP391.4
- DOI:
-
10.3969/j.issn.1673-4785.2009.06.012
- 文献标志码:
-
A
- 摘要:
-
目标跟踪是计算机视觉和图像处理的一个重点课题,在视频监控、机器人视觉导航以及智能交通控制中具有广泛的应用前景.通过粒子滤波技术,研究了如何整合颜色特征、前景信息和积分图运算等技术实现视频目标跟踪的粒子滤波算法.在对目标进行分割中采用了混合高斯背景建模方法;同时结合积分直方图的计算方法对颜色特征进行分段统计及相互遮挡的判断,实现基于粒子滤波的目标跟踪算法的优化,解决跟踪中诸如遮挡、光照变化、背景干扰、尺寸变化等难以解决的问题.实验结果表明提出的方法达到了预期目标.
- Abstract:
-
Tracking moving objects across video sequences is a key subject in computer vision and image processing. It is required for fields as diverse as video surveillance, robotic vision, navigation, and intelligent traffic control. After studying particle filter technology, a moving object tracking algorithm based on a particle filter was developed, integrating foreground information, color features and an integral image method. Gaussian mixture models (GMM) were employed for object foreground segmentation. By using an integral histogram algorithm, data for the color features in different sections were derived. Based on the feature’s integral histogram, occlusion could then be analyzed and judged. The method presented can resolve occlusion caused by objects appearing among obstacles or with other objects, and to some extent overcome problems caused by changes in illumination, background and apparent size. Experimental results when tracking moving objects agreed well with predicted results.
更新日期/Last Update:
2010-02-22