拓展目标杂波概率假设密度估计

Estimation of Clutter Probability Hypothesis Density in Extended Objects

针对拓展目标概率假设密度滤波器中的未知杂波概率假设密度,提出了杂波概率假设密度估计算法。算法利用有限混合模型极大后验估计杂波概率假设密度,取混合权重的熵分布作为混合参数的先验分布;在渐进假设条件下,利用拉格朗日乘子推导了混合权重的递进估计公式;在混合权重递进估计过程中,通过混合权重置零操作来实现对有限混合模型中混合分量的删减。以二维场景为例对算法进行了仿真实验,结果表明:在拓展目标概率假设密度滤波器高斯混合实现的框架内,所提杂波概率假设密度估计算法的跟踪性能接近真实杂波概率假设密度时的跟踪性能。

An algorithm to estimate the clutter probability hypothesis density is proposed to deal with unknown clutter probability hypothesis densities in extended object probability hypothesis density filter. A finite mixture model is applied in estimating clutter probability hypothesis density with maximum a posterior estimation, and the entropy distribution of mixed weights is regarded as the prior distribution of mixing parameters. A recursive estimation formula of mixed weights is derived by using Lagrange multiplier under an approximation assumption. Mixture components of the finite mixture model are pruned by setting corresponding mixed weights to zeros in the recursive estimation procedure. Simulation results in a two-dimensional scenario show that the tracking performance of the proposed algorithm is close to the ground truth in Gaussian mixture implementation of extended object probability hypothesis density filter.