摘要
目标跟踪是监控系统的关键技术之一。Mean-Shift作为高效的模式匹配算法,已经成功地应用在对实时性要求较高的目标跟踪系统中,但是传统的Mean-Shift跟踪算法未能有效地解决跟踪窗自适应和目标特征更新问题,无法实现对目标的长时间跟踪。论文提出了卡尔曼滤波、局部目标检测和Mean-Shift有机结合的目标自适应跟踪方法,能有效的解决跟踪窗自适应和跟踪目标的特征更新问题,并有较强的抗遮挡的能力,提高跟踪的稳健性。并通过跟踪实验对比验证了算法的有效性。
Object tracking is one of the key technologies in multi-camera surveillance system.As a high-performance pattern-matching algorithm, Mean-Shift has been successfully applied to many real-time tracking systems.But classic Mean-Shift based tracking algorithm can not effectively solve the problem of tracking window adaptivity and characteristics update.So,it can not track an object for a long time.In this paper,an adaptive tracking method is proposed which combines Kalman filter,local object detection and Mean-Shift.This method can be an effective solution to tracking window adaptivity and characteristics update,and also has a strong anti-blocked ability.Some experiments are applied to verify the effecavity of the algorithm.
出处
《微型电脑应用》
2009年第5期14-16,4,共3页
Microcomputer Applications
