[1]陈丽萍,王铭羽,杨文柱,等.基于改进核相关滤波的长时目标跟踪算法[J].郑州大学学报(工学版),2020,41(03):27-32.[doi:10.13705/j.issn.1671-6833.2019.02.001]
 Chen Liping,Wang Mingyu,Yang Wenzhu,et al.Long-term Object Tracking Based on Improved Kernelized Correlation Filters[J].Journal of Zhengzhou University (Engineering Science),2020,41(03):27-32.[doi:10.13705/j.issn.1671-6833.2019.02.001]
点击复制

基于改进核相关滤波的长时目标跟踪算法()
分享到:

《郑州大学学报(工学版)》[ISSN:1671-6833/CN:41-1339/T]

卷:
41卷
期数:
2020年03期
页码:
27-32
栏目:
出版日期:
2020-07-29

文章信息/Info

Title:
Long-term Object Tracking Based on Improved Kernelized Correlation Filters
作者:
陈丽萍王铭羽杨文柱王思乐陈向阳
河北大学网络空间安全与计算机学院
Author(s):
Chen LipingWang MingyuYang WenzhuWang SileChen Xiangyang
School of Cyberspace Security and Computer, Hebei University
关键词:
长时目标跟踪模板漂移核相关滤波置信度
Keywords:
Long-term target trackingtemplate driftKernel correlation filteringConfidence
DOI:
10.13705/j.issn.1671-6833.2019.02.001
文献标志码:
A
摘要:
针对长时间目标跟踪过程中因快速移动和物体遮挡而导致的模板漂移问题,提出了基于改进核相关滤波的长时目标跟踪算法(LKCF)。该算法以核相关滤波算法(KCF)为跟踪框架,采用了高置信度的模版更新策略防止模版破坏,并设置了一个目标重检机制用于在跟踪失败时找回丢失的目标。实验结果表明所提出的算法不仅有效避免了模版漂移问题,并且可以长期稳定的跟踪目标。
Abstract:
In order to solve the problem of template drift caused by fast moving and object occlusion in long term target tracking, a Long-term Kernelized Correlation Filter (LKCF) based on improved Kern el Correlation Filter (KCF) was proposed. In this paper, we used the kernel correlation filter algorithm(KCF) as the tracking framework and adopted a highly reliability template update strategy to prevent template destruction. Furthermore, we constructed a conditional target re-detection mechanism to restore the false model caused during tracking. Experimental results indicate that the proposed algorithm not only avoid the problem of template drift effectively but also can track targets steadily for a long time.
更新日期/Last Update: 2020-07-29