箱粒子广义标签多伯努利滤波的目标跟踪算法

发布时间：2018-06-06 04:22

本文选题：目标跟踪 + 随机有限集　；参考：《西安交通大学学报》2017年10期

【摘要】：针对序列蒙特卡罗广义标签多伯努利滤波(SMC-GLMB)算法计算效率低、实时性差的问题,提出了箱粒子广义标签多伯努利滤波的目标跟踪(Box-GLMB)算法。该算法使用带标签的随机有限集描述多目标的状态,包括目标的位置和速度,并且对每个目标用互不相同的标签进行区分;然后利用箱粒子滤波算法近似单目标状态的概率密度,即用一组带权值的均匀分布拟合单目标状态概率密度;最后通过广义标签多伯努利滤波对多目标状态的概率密度进行预测与更新,从多目标状态后验概率密度中估计单目标的位置与速度,根据目标的标签可以实现航迹跟踪。BoxGLMB算法结合了箱粒子滤波与GLMB算法的优势,能够跟踪目标航迹,同时提高计算效率。仿真结果表明,Box-GLMB算法可以有效估计目标状态以及跟踪目标航迹,相比于SMC-GLMB算法,计算效率提升了62%。
[Abstract]:Aiming at the problems of low computational efficiency and poor real-time performance of the sequential Monte Carlo generalized label multiple Bernoulli filter (SMC-GLMB), a Box-GLMB-based target tracking algorithm based on the box particle generalized label multiple Bernoulli filter is proposed. The algorithm uses a random finite set with labels to describe the state of multiple targets, including the position and speed of the targets, and distinguishes each target with different labels. Then the box particle filter algorithm is used to approximate the probability density of a single target state, that is, a set of weighted uniform distribution is used to fit the probability density of a single target state. Finally, the probability density of multi-target state is predicted and updated by generalized label multi-Bernoulli filter, and the position and velocity of single target are estimated from the posteriori probability density of multi-objective state. According to the tag of the target, track tracking. BoxGLMB algorithm combines the advantages of box particle filter and GLMB algorithm, can track the track of the target and improve the computational efficiency. Simulation results show that the Box-GLMB algorithm can effectively estimate the state of the target and track the target. Compared with the SMC-GLMB algorithm, the computational efficiency is improved 62%.
【作者单位】：西安电子科技大学电子工程学院;
【基金】：国家自然科学基金资助项目(61372003)
【分类号】：TN713

【相似文献】