基于均方误差的多目标概率假设密度滤波器

The Mean Square Error-based Multi-target Probability Hypothesis Density Filter

在噪声、杂波和漏检等因素干扰的目标跟踪下,标准概率假设密度滤波器的目标状态估计精度及其计算效率低,难以满足目标跟踪系统的要求。文章提出一种基于均方误差的多目标概率假设密度滤波器,主要包括目标量测和杂波划分以及目标分量选择与更新策略。目标量测和杂波划分策略识别源于真实目标的量测以构建目标量测集;目标分量选择与更新策略通过似然函数和卡尔曼增益选择和更新目标。结果表明,相对复杂跟踪场景下本文算法不仅具有较高的目标状态估计精度,而且具有相对较高的计算效率。

Under the target tracking of the interference uncertain factors such as noise,clutter and missed detection,the estimate accuracy of the target states and computational efficiency of the standard probability hypothesis density filter is low,which is difficult to meet the requirements of the target tracking system.In this paper,a multi-target probability hypothesis density filter based on mean square error is proposed,which mainly includes target measurement and clutter division scheme and target component selection and update scheme.Using the target measurement and clutter division scheme,the measurements originating from real targets are identified to construct target measurement set.The target component selection and update strategy selects and updates the target by the likelihood function and Kalman gain,so as to achieve the purpose of accurately updating the target intensity and optimizing the target component in the posterior intensity.Results show that the proposed algorithm has not only the high target state estimate accuracy,but also relatively high computational efficiency in the relative complex tracking scenarios.