多扩展目标的高斯混合概率假设密度滤波器

发布时间：2018-07-29 19:13

【摘要】：针对多扩展目标跟踪中状态信息难以估计的问题,提出了一种可以估计扩展目标运动状态和形状信息的多扩展目标高斯混合概率假设密度(RHM-GMPHD)滤波器。首先利用描述凸星形扩展目标量测源分布的随机超曲面模型和传感器量测方程,建立扩展目标运动状态及形状信息与量测之间关系的伪量测函数;然后结合扩展目标状态预报信息,推导了扩展目标状态更新方程,递推地对扩展目标运动状态及形状信息进行估计跟踪。此外,还建立了Jaccard距离来度量RHMGMPHD滤波器对目标形状的估计性能。与联合概率数据关联(JPDA)滤波器和GMPHD滤波器相比,RHM-GMPHD滤波器不仅可以估计凸星形扩展目标的形状信息,并能有效提高对目标数和运动状态的估计精度。仿真实验表明,RHM-GMPHD滤波器对质心估计的均方根误差分别约为JPDA和GMPHD滤波器的1/3和1/2,对目标数的估计接近真实值,对形状估计的Jaccard距离一般小于0.2。
[Abstract]:In order to solve the problem that state information is difficult to estimate in multi-extended target tracking, a hybrid probability assumption density (RHM-GMPHD) filter for multi-extended target Gao Si is proposed, which can estimate the state and shape information of extended target. Firstly, using the random hypersurface model and sensor measurement equation to describe the distribution of the convex star extended target source, the pseudo measurement function of the relation between the moving state and shape information of the extended target and the measurement is established. Then, the extended target state update equation is derived by combining the extended target state prediction information, and the state and shape information of the extended target are estimated and tracked recursively. In addition, the Jaccard distance is established to measure the performance of the RHMGMPHD filter to estimate the shape of the target. Compared with the joint probabilistic data association (JPDA) filter and GMPHD filter, the RHM-GMPHD filter can not only estimate the shape information of the convex star extended target, but also improve the estimation accuracy of the target number and motion state. The simulation results show that the RHM-GMPHD filter has a root mean square error of about 1 / 3 and 1 / 2 of the JPDA and GMPHD filters, respectively. The estimation of the target number is close to the real value, and the Jaccard distance for shape estimation is generally less than 0.2.
【作者单位】：西安交通大学电子与信息工程学院;西安邮电大学电子工程学院;
【基金】：国家自然科学基金资助项目(61203220,61221063,61074176) 国家“973计划”资助项目(2013CB329405)
【分类号】：TN713;TN953

【参考文献】