卡尔曼预测器在目标相关跟踪中的应用

在目标跟踪系统中,特别是在复杂背景情况下对地面目标的跟踪中,相关跟踪算法是常用的一种算法.但问题是传统相关算法采用全局搜索的方法,使得计算量相当大,不易实时实现,而且当发生目标局部遮挡时,目标容易丢失.为解决这个问题,该文提出一种基于卡尔曼预测器的目标相关跟踪的方法,充分利用卡尔曼预测器的预测功能来预测下一帧目标可能出现的区域,然后在较小的预测区域中进行相关匹配运算,找到最佳相关匹配点,从而使目标相关跟踪更具主动性.实验中用传统算法和本文提出的算法对高速行驶的坦克进行跟踪时,传统算法容易跑飞,而该文算法不受遮挡干扰,始终稳定跟踪且耗时大幅减少.因此该文算法能够较好地克服传统相关算法中存在的计算量大和易受遮挡干扰的缺点.

Kalman Filter in Target Tracking

The correlation is an usual method in target tracking system,especially in the complex background.But there is a problem in traditional matching method that the method searches the target in the whole area,so it can't be realized in real time because of large amounts of calculation.And when the target is partial shaded,the target will be lost on tracking.To solve the problem,this paper presents a new approach based on the Kalman Filter to realize target tracking.And this algorithm uses the forecast function of Kalman Filter to make target tracking more active.In experiment,the traditional method always lose the tank when the tank is at a high speed,and the method of the paper can steadily track it.So the method of the paper can well overcome the main disadvantages of large amounts of calculation,and more steady in case when target is partial shaded.