摘要
为解决相关滤波一类跟踪器因目标被遮掩等原因引起的跟踪目标丢失问题,提出一种包含遮掩感知和目标重检测机制的相关滤波器改进跟踪方法。采用遮掩感知模块对目标跟踪模块跟踪的结果做遮掩事件评估,在发生遮掩等严重影响跟踪结果的情形时,采用目标重检测模块重新检测原跟踪目标,该目标重检测采用基于颜色特征的逐像素目标置信积分图来确定备选目标。对跟踪质量不可靠且未重检测到可靠目标的视频帧,不进行跟踪模型的在线更新。实验结果表明,该算法可以有效避免因遮掩等引起的跟踪目标丢失和模型漂移问题,跟踪性能和几个主流的相关滤波类跟踪器相比有明显改善。
To solve the problem of target loss due to occlusion for a variety of correlation filter based trackers,an improved trac-king method was proposed based on occlusion awareness and target re-detection mechanism,in which the occlusion awareness module was used to evaluate whether the tracked object was occluded,or whether the tracking result was reliable.As the events such as object occlusion that resulted in tracking failure occur,the object re-detection module was triggered to re-detect the original tracking target based on integral map of pixel-wise object confidence by color information.When the tracking quality was unreliable and no reliable object was re-detected,the tracking model was not updated.Experimental results show that the proposed algorithm can effectively avoid the problem of the correlation filter tracker’s variants,tracking object loss and model drift caused by occlusion,its tracking performance is obviously improved compared with that of several state-of-the-arts correlation filter tracker’s variants.
作者
李中科
万长胜
LI Zhong-ke;WAN Chang-sheng(School of Computer and Software,Nanjing Institute of Industry Technology,Nanjing 210046,China;School of Information Science and Engineering,Southeast University,Nanjing 210096,China)
出处
《计算机工程与设计》
北大核心
2019年第2期567-572,595,共7页
Computer Engineering and Design
基金
江苏省自然科学基金项目(BK20161099)
2017年江苏省高校优秀科技创新团队基金项目(SJCXTD-03工业大数据应用技术)
南京工业职业技术学院科研基金项目(YK15-04-01)
