摘要
传统基于颜色表示的在线目标跟踪方法,倾向于跟踪与目标外观相似的区域,会因为尺度变化而导致漂移。针对该问题,结合干扰感知模型与背景对象模型,提出一种基于颜色表示的目标跟踪方法。通过干扰感知模型抑制干扰区域,利用背景对象模型将目标对象从周围背景中区分出来,并结合自适应尺度估计方法进行目标跟踪。实验结果表明,与STC和RVT跟踪方法相比,该方法在精度和鲁棒性方面表现更好。
In an online target tracking based on color representation,in order to solve target tracking,it tends to track the area that is similar to the tracking target, and the problem of drift caused by the change of target scale. In this paper,a target tracking method is obtained by combining the interference perception model and the background object model.The interference model can effectively suppress the interference area, and the background object model can distinguish the target object from the surrounding background. Furthermore,the adaptive scale estimation method is added. The experimental results show that,compared with STC tracking algorithm and RVT tracking method,the proposed method has better performance in precision and robustness.
作者
张志凡
谢世朋
傅鹏
ZHANG Zhifan;XIE Shipeng;FU Peng(School of Telecommunications and Information Engineering,Nanjing University of Posts and Telecommunications,Nanjing 210003,China)
出处
《计算机工程》
CAS
CSCD
北大核心
2018年第7期225-229,共5页
Computer Engineering
基金
国家自然科学基金(11547155)
教育部-中国移动科研基金(MCM20150504)
江苏省科技重点研发计划-产业前瞻与共性关键技术项目(BE2016001-4)
南京邮电大学科研基金(NY214026
NY217035)
关键词
对象模型
尺度自适应
干扰感知模型
背景对象模型
颜色直方图
object model
scale adaptation
interference perception model
object-background model
color histogram
