基于箱粒子滤波的混合标签多伯努利跟踪算法

针对标签多伯努利滤波器在目标处于近邻或目标量测与轨迹关联模糊情况下,更新步中由于近似产生信息丢失,导致跟踪效果下降的问题,引入区间分析技术,结合标签多伯努利滤波器及广义标签多伯努利滤波器各自的优势,提出一种箱粒子滤波下的混合标签多伯努利跟踪算法.建立两种滤波器的参数模型,通过Kullback Leibler散度和熵两项评定标准在两种滤波器间进行切换,在特殊环境中使用广义标签多伯努利滤波器提高跟踪性能,在其他环境中使用标签多伯努利滤波器近似降低算法的复杂度,提高运算效率;同时基于箱粒子滤波实现混合标签多伯努利算法.仿真实验表明,在特定环境中,与原有滤波算法相比,所提出的改进算法在保证计算效率的同时,可提高跟踪的精确度及稳定性.

A hybrid labeled multi-Bernoulli tracking algorithm based on box particle filter

In view of the problems that the labeled multi-Bernoulli has a decline tracking effect when the targets are close or track-to-measurement association is ambiguous because of the approximate information loss in update step, the interval analysis technology is introduced. Combined with respective advantages of the generalized labeled multi-Bernoulli (GLMB) and labeled multi-Bernoulli (LMB), a hybrid labeled multi-Bernoulli tracking algorithm based on box particle filter(Box-HLMB) is proposed. The GLMB and LMB parameter sets are established. By switch between the GLMB and LMB based on the Kullback Leibler divergence and entropy evaluation criteria, the GLMB is used in the critical environment to improve the tracking performance, LMB approximation is used in other environment to improve the efficiency of operation. The hybrid labeled multi-Bernoulli algorithm is implemented based on box particle filter. The simulation results show that compared with the GLMB and LMB filtering algorithms, the improved algorithm can ensure the computational efficiency, as well as improving the accuracy and stability of the tracking performance.