的个人主页 http://faculty.nuaa.edu.cn/ghj2/zh_CN/index.htm
点击次数:
所属单位:民航学院
发表刊物:Zhejiang Daxue Xuebao (Gongxue Ban)
摘要:A fast object tracking algorithm via kernel collaborative presentation was proposed to solve the problem of single object tracking in complex scenes. In the framework of particle filter, square matrix was introduced into the dictionary matrix to model the occlusion appeared in the object tracking. Then the dictionary matrix and candidates were mapped to the high dimensional space, separately; the linear representation model between the candidates and dictionary matrix in high space was established. L2 regularization was used to reduce the requirement of the traditional method for the sparsity of coefficients, which can effectively reduce the computational complexity of the key steps. Experimental results show that the proposed algorithm can effectively overcome the influence of occlusion, illumination change, scale change and motion blur, with higher average overlap rate and lower average center location error. © 2017, Zhejiang University Press. All right reserved.
ISSN号:1008-973X
是否译文:否
发表时间:2017-02-01
合写作者:Wang, Hai-Jun,Zhang, Sheng-Yan
通讯作者:Wang, Hai-Jun,葛红娟