基于多模型的联合目标跟踪与分类算法

A Joint Target Tracking and Classification Algorithm Based on Global Multi-Model

针对现有的联合目标跟踪与分类算法不具备封闭解析形式、计算量大且缺乏模块化结构等特点,将预知的各类目标的多种模型联合起来,组成一个多模型集合,并在运动学传感器和属性传感器观测过程相互独立的前提下,对目标状态概率密度函数和目标类别概率质量函数同时进行贝叶斯推理,得出一种基于多模型的联合目标跟踪与分类算法。该算法由卡尔曼联合多模型滤波器和贝叶斯分类器组成,实现了跟踪器与分类器的模块化,提高了跟踪和分类性能,具有封闭解析形式,计算量较小,适合工程实时应用。通过仿真,证实了该算法的有效性。

In view of the high computational complexity of the existing joint target tracking and classification （JTC） algorithm, which has neither closed form nor modular structure, we united the models of all predicted target types to form a global multi-model set. Then, we proposed a joint target tracking and classification algorithm based on global multiple-model（GMM-JTC） by applying Bayes＇ rule to the target state probability density function and target class probability mass function simultaneously under the assumption that the kinematic and attribute measurement processes are conditional independent. The GMM-JTC algorithm, which consists of a Kalman global multiple-model filter and a Bayesian classifier, has a closed form with a modularized structure, together with a lower computational complexity. It＇s more suitable for real-time applications. The simulation results confirm the effectiveness of the proposed GMM-JTC algorithm.