基于蚁群算法的多目标跟踪方法

New method for multi-target tracking using ant algorithm

提出了一种新的基于蚁群算法的多目标跟踪方法。方法采用蚁群算法实现多目标跟踪中的数据关联,首先将多目标跟踪中的数据关联问题表示为具有约束条件的优化问题,用蚁群算法对该优化问题求解,得到的解即为最优关联。为验证该算法的有效性,在两种状态估计方法EKF(extended Kalman filter)和SIS(sequential im-portance sampling)的基础上进行了多目标跟踪实验,并且与传统的NN(nearest neighbor)方法进行了比较。在与SIS框架结合时,算法中采样粒子包括状态矢量和关联矢量,状态矢量通过序贯重要性重采样获得,关联矢量通过蚁群算法求得。实验结果表明,将蚁群算法融合进SIS算法进行多目标跟踪是有效的。

A method based on ACA（ant colony algorithm） is proposed for data association for multi-target tracking. Firstly, the data association problem is represented as a formulation of combinational optimization, then, ant algorithm is used to solve the optimization problem. SIS（sequential importance sampling） is introduced to combine with the proposed method to complete the multi-target tracking, where state vector is obtained by sampling from a distribution and association vector by ant algorithm with probability one, which decrease the uncertainty of the probabilistic method. The proposed method combined with EKF（extended Kalman filter） and SIS is compared with NN（nearest neighbor） respectively. Simulation results show that the proposed method is an effective way in the field of multi-target tracking.