基于回归分析的多传感器时间配准跟踪方法

Method of multi-sensor time registration and target tracking based on regression analysis

为提高大数据量情况下目标跟踪数据处理效率,将各传感器获取的目标测量值在时间轴上依次排列,利用成组序贯数据回归分析的目标跟踪方法,实现时间配准与目标跟踪的同步处理.首先对滤波曲线两点回归分析,逐一预测各观测点到达时刻的置信区间,筛选关联点;然后按系统处理周期分组,对关联点进行成组观测数据回归分析,计算出观测融合点,再用观测融合点更新滤波器新息;最后在Kalman滤波器框架下,实时更新运动方程,递推完成目标连续跟踪.仿真结果表明,本文时间配准跟踪方法能同步完成时间配准和目标跟踪,在节约资源的同时提高系统跟踪精度.

In order to improve the efficiency of target tracking data processing in the case of large data volume, the target measurement values obtained by each sensor are sequentially arranged on the time axis, and the synchronous processing of time alignment and target tracking is realized by using the target tracking method of group sequential data regression analysis. Firstly, this paper predicts the confidence interval of arrival time of each observation point by two-point regression analysis of the filter curve, with the correlation points screened, and then groups the data according to the system processing cycle, analyzing the observation data by regression analysis to calculate the observation fusion point, and using the observation fusion point to update the new information of the filter. Finally, in the framework of Kalman filter, the paper updates the motion equation in real time, and recursively completes the continuous tracking of the target. Simulation results show that the proposed time registration tracking method can synchronize time alignment and target tracking, and also improve the tracking accuracy of the system while saving resources.