北京航空航天大学学报 ›› 2015, Vol. 41 ›› Issue (3): 559-566.doi: 10.13700/j.bh.1001-5965.2014.0455

• 论文 • 上一篇    

基于L2范数最小化联合模型的目标跟踪算法

王蒙, 吴毅, 邓健康, 刘青山   

  1. 南京信息工程大学 信息与控制学院, 南京 210044
  • 收稿日期:2014-04-28 出版日期:2015-03-20 发布日期:2015-04-02
  • 通讯作者: 刘青山(1975—),男,安徽合肥人,教授,qsliu@nuist.edu.cn,主要研究方向为图像处理与模式识别. E-mail:qsliu@nuist.edu.cn
  • 作者简介:王蒙(1990—),女,江苏盐城人,硕士生,wangmeng2008ji@163.com
  • 基金资助:
    国家自然科学基金资助项目(61370036,61005027); 江苏省自然科学基金资助项目(201204234,201210296)

Object tracking based on the joint model using L2-norm minimization

WANG Meng, WU Yi, DENG Jiankang, LIU Qingshan   

  1. School of Information & Control, Nanjing University of Information Science and Technology, Nanjing 210044, China
  • Received:2014-04-28 Online:2015-03-20 Published:2015-04-02

摘要: 为了解决稀疏表示的跟踪算法的计算代价比较大,且目标的表观由于多种原因会发生变化的问题,提出了一种在贝叶斯推理框架下,建立结合基于全局模板的判别式模型和基于局部描述子的生成式模型的联合模型,通过L2范数最小化进行求解的目标跟踪方法.在跟踪过程中,适时地更新判别式模型中的正负模板和生成式模型中模板的系数向量,使模板具有很强的适应性和判别性.实验结果表明,与其他典型的算法相比,该算法对于光照变化、尺度变化、遮挡、旋转等情况具有较强的鲁棒性.

关键词: 目标跟踪, L2范数最小化, 判别式模型, 生成式模型, 子空间

Abstract: The computational cost of the tracking algorithm based on the sparse representation is so much large, at the same time, the target apparence changes on account of a variety of reasons,which makes the object tracking process complicated and time consuming. A joint model is reasonably proposed by combining the global template based on the discriminant model and the generation model based on the local descriptor, properly solved by the L2-norm minimization solution in a bayesian inference framework, which is proved to be effective and efficient. In the process of the object tracking process, the plus template and the minus template of the discriminant model and the coefficient vector of the generative model are timely updated so as to have a strong adaptability and robust discrimination. The experimental results finally show that compared with other typical algorithms, the proposed algorithm has stronger robustness in the case of illumination, scale changes, shelter, rotation and so on.

Key words: object tracking, L2-norm minimization, discriminative model, generative model, subspace

中图分类号: 


版权所有 © 《北京航空航天大学学报》编辑部
通讯地址:北京市海淀区学院路37号 北京航空航天大学学报编辑部 邮编:100191 E-mail:jbuaa@buaa.edu.cn
本系统由北京玛格泰克科技发展有限公司设计开发