群分裂是群目标的重要运动形式,基本的广义标签多贝努利(GLMB)滤波器计算复杂,且只对存活和新生目标建模,不能有效描述群分裂运动。针对此问题,在δ-GLMB 滤波算法的基础上建立群分裂模型,通过最小化Kullback-Leibler 散度(KL-DIV)的方法近似目标真实密度分布,推导了一种考虑群分裂的GGIW鄄啄鄄GLMB 滤波算法。仿真实验表明,该算法能够较好地跟踪群分裂运动。

Splitting is an motion manner of group target. The original generalized labeled multi-Bernoulli (GLMB) filter, which is complex in calculation, only the surviving and new born targets are modeled in the filter, and the group spawning could not be handled. Aimed at this, one group spawning model is introduced in Air Force Engineering University -GLMB filter, and the Gamma Gaussian Inverse Wishart Air Force Engineering University -GLMB (GGIW- Air Force Engineering University-GLMB) filter with group spawning considered is derived with the true density of the group approximated by Kullback-Leibler divergence (KL-DIV). Simulation shows, the proposed algorithm does improve the performance for group spawning events.

TN957.52