基于MM-GGIW-PMBM滤波的扩展目标跟踪算法

针对多机动扩展目标跟踪问题，将交互式多模型的思想引入泊松多伯努利混合滤波(PMBM)算法中，提出了一种多模型的伽马高斯逆威夏特-泊松多伯努利混合滤波(MM-GGIW-PMBM)算法。该算法融合多种运动模型，通过模型的交互实现对机动扩展目标扩展状态和质心状态的混合估计预测；通过引入强跟踪滤波(STF)中的渐消因子修正预测之后GGIW分量中的协方差矩阵，防止发生跟踪模型失配的现象；在PMBM更新阶段扩展目标外形和质心估计完成的基础上，利用似然函数完成模型概率的更新。仿真实验结果表明:MM-GGIW-PMBM算法能够对多机动扩展目标的数量和状态进行有效的估计。 

Extended target tracking algorithm based on MM-GGIW-PMBM filter

To address the problem of multiple maneuvering extended target tracking, this paper introduces the concept of interactive multiple models into the Poisson multi-Bernoulli mixture filtering (PMBM) algorithm, and proposes a multi-model algorithm with Gamma Gaussian inverse Wishart and PMBM (MM-GGIW-PMBM). Firstly, the algorithm integrates multiple motion models and realizes the hybrid estimation and prediction of the extended state of the maneuvering target and the centroid state through the interaction of the models. Secondly, the covariance matrix in the predicted GGIW components is modified by introducing the fading factor into the strong tracking filter (STF) to prevent the tracking model mismatch. Finally, the target shape is expanded in the PMBM update stage based on the completion of centroid estimation, and the likelihood function is used to update the model probability. The simulation shows that the proposed algorithm can effectively estimate the number and state of multiple maneuvering extended targets.