基于高斯粒子CPHD滤波的多目标检测前跟踪算法

发布时间：2018-06-15 15:48

本文选题：检测前跟踪 + 势概率假设密度　；参考：《控制与决策》2017年11期

【摘要】：针对未知目标数条件下多弱小目标检测前跟踪(TBD)算法鲁棒性较低、运算量较大等问题,提出一种基于高斯粒子势概率假设密度(CPHD)滤波的多目标检测前跟踪算法.运用高斯函数近似目标状态的后验概率密度,采取粒子滤波的方法迭代更新CPHD中各高斯项的均值与协方差,无需重采样,避免了粒子退化和采样枯竭等问题;同时结合检测前跟踪算法的实际情况,得出粒子权值的更新表达式.仿真实验表明,与现有算法相比,所提出算法在降低复杂度的同时,可以更为可靠地传递目标势分布信息,从而提高多弱小目标数目和状态估计的准确性和稳定性.
[Abstract]:In order to solve the problems of low robustness and large computational complexity in multi-dim target pre-tracking algorithm with unknown number of targets, a multi-target detection pre-tracking algorithm based on Gao Si particle potential probability assumption density (Gao Si) filter is proposed. Using the Gao Si function to approximate the posterior probability density of the target state, the particle filter method is used to iteratively update the mean and covariance of each Gao Si term in the CPHD without re-sampling, and the problems of particle degradation and sampling depletion are avoided. At the same time, according to the actual situation of the tracking algorithm before detection, the update expression of particle weight is obtained. Simulation results show that compared with the existing algorithms, the proposed algorithm can transfer the target potential distribution information more reliably while reducing the complexity, thus improving the accuracy and stability of the number of small and weak targets and the state estimation.
【作者单位】：空军工程大学信息与导航学院;95806部队;
【基金】：国家自然科学基金项目(61571458) 陕西省自然科学基金项目(2011JM8023)
【分类号】：TN713

【相似文献】