基于标签随机有限集滤波器的多扩展目标跟踪算法

针对现有随机有限集(random finite set, RFS)扩展目标滤波器不能输出航迹的问题,提出了基于标签RFS滤波器的多扩展目标跟踪算法。该算法首先采用随机超曲面模型将目标建模为星-凸扩展形态,然后利用标签策略表征集合中的离散元素,结合基于延迟逻辑的多假设跟踪理论,采用N 次回扫策略对多帧量测进行平滑处理。仿真实验结果表明,该算法可以在目标跟踪过程中形成完整航迹并对目标扩展形态进行有效估计,特别是在低信噪比探测场景中,所提算法跟踪精度明显优于传统RFS滤波算法,进一步提高了滤波器的稳定性和有效性。

Multiple extended target tracking algorithm based on labeledrandom finite set filter

A multiple extended target tracking algorithm based on labeled random finite set (RFS) filter is proposed to solve the problem that the existing algorithm based on the RFSs cannot output the target kinematic track. The proposed algorithm adopts the random hypersurface to model the extended target as star-convex extension, and then puts the elements in the sets labeled. Combining with the delay logic theory of multiple hypothesis tracking, we adopt the backtracking strategy to smooth multiframe measurements. Experimental results show that the proposed algorithm is able to output complete track and effectively estimate the kinematic state and target extension when tracking the target, especially in the low signal-to-noise ratio scenario, the tracking accuracy of the proposed algorithm is better than that of the traditional extended target filter based on the RFS theory, so that the stability and effectiveness of the filter can be improved.