论文部分内容阅读
针对伽马高斯逆威夏特-概率假设密度(GGIW-CPHD)滤波算法跟踪机动群目标误差较大的问题,提出基于最适高斯近似(BFG)和强跟踪的多模型GGIW-CPHD滤波的群跟踪算法.首先,在对群目标量测分割的基础上,采用BFG方法实现CPHD预测阶段的多模型融合.其次,利用强跟踪滤波(STF)中的渐消因子来修正GGIW分量的预测协方差矩阵.最后,在CPHD更新阶段完成群质心和扩展状态估计的基础上,利用多个模型对应的似然函数完成模型概率的更新.实验结果表明:所提算法能够在GGIW-CPHD框架下实现多个模型的交互,有效降低机动阶段时群目标的状态估计误差,并能有效处理群目标的合并和衍生情况.
In view of GGIW-CPHD filtering algorithm tracking the large target error of maneuvering group, a multi-model GGIW-CPHD filtering group based on the best Gaussian approximation (BFG) and strong tracking is proposed Tracking algorithm.Firstly, BFG method is used to realize the multi-model fusion of CPHD prediction stage based on the group target measurement and segmentation.Secondly, the fading factor in strong tracking filter (STF) is used to correct the prediction covariance of GGIW components Matrix.Finally, updating the model probabilities using the likelihood function of multiple models based on the completion of the population centroid and the extended state estimation in the CPHD update phase.The experimental results show that the proposed algorithm can achieve more in the framework of GGIW-CPHD The model interaction can effectively reduce the state estimation error of group targets in the maneuver phase and can effectively deal with the mergence and derivation of group targets.