基于STF和加权改进的群目标跟踪算法

为了进一步提高群目标交互多模型跟踪算法的估计性能，提出一种改进的群跟踪算法.首先，通过采用模型转换概率的自适应算法，优化模型与目标运动模式的实时匹配.并通过引入强跟踪滤波（STF，Strong Tracking Filter）中的渐消因子，提高机动阶段时的群质心的状态估计精度.其次，分别利用概率加权法和标量加权法完成群质心状态和扩展状态的融合估计.最后在变分贝叶斯滤波的基础上，建立完整的跟踪算法流程.仿真实验结果表明，该方法不仅能够提高群质心状态和扩展状态的估计精度，还能有效降低机动阶段时的峰值误差.

Group targets tracking algorithm based on strong tracking filter and improved weighted method

To improve the estimation performance of the existing interactive multiple models tracking algorithm for group targets, an improved group tracking algorithm was proposed. Firstly, by using the adaptive algorithm of model transition probability, the optimization of real-time matching for tracking models with the actual motion pattern was performed. And a fading factor of strong tracking filter was used to improve the estimation accuracy of the centroid state in the maneuvering stage. Then the fusion estimation of centroid state and extension state were implemented by using the probability weighted method and the scalar coefficients weighted method, respectively. Lastly, the implementation steps of the new tracking algorithm were presented in detail, which were based on variational Bayesian filtering algorithm. The computer simulations show that the estimation accuracy of the centroid state and extension state is improved in the new algorithm, and this algorithm can reduce a great deal of peak error in the maneuvering stage.