摘要
传统均值漂移(Mean-shift)算法是一种半自动的跟踪方法,在目标被遮挡的情况下无法进行有效的跟踪。结合背景减除法提出了一种新的抗遮挡跟踪方法。利用背景减除法在初始帧确定目标的运动区域,得到Mean-shift的初始化参数;在跟踪过程中提出了一种目标遮挡因子作为目标被遮挡程度的判断依据,并根据目标被遮挡的程度提出相应的解决策略。实验结果表明该方法克服了传统Mean-shift算法需要人为定位的缺点,且在全遮挡的情况下仍可以正确地跟踪目标。
Traditional Mean-shift is a semi-automatic tracking algorithm which could not work well when the target is occluded.A new anti-occlusion method in combination of the background subtraction is proposed in this paper.Firstly,by using the background subtraction,the moving region of object is confirmed in initial frame,thus to obtain the initialized parameter of the Mean-shift algorithm.In the tracking process,a target-occlusion coefficient is proposed,and relevant strategies are suggested according to the being-occluded degree.Experimental results show that this method overcomes the shortcoming of artificial orientation in traditional Mean-shift,and could correctly track the targets in complete occlusion.
出处
《信息安全与通信保密》
2010年第9期65-67,共3页
Information Security and Communications Privacy
关键词
传统均值漂移
背景减除法
遮挡因子
mean-shift
background subtraction
occlusion coefficient
