分类信息辅助的多目标跟踪方法研究

多目标跟踪系统的关键技术之一是航迹关联。当传感器能同时得到目标分类信息和运动信息时，本文提出结合分类信息的综合概率数据关联算法，把目标不同信息相结合来提高关联效果。它通过分类混淆矩阵确定分类信息似然函数。再用谊函数调整传统的只利用运动信息的似然函数。使分类信息有效辅助综合概率数据关联．在杂波环境对多个邻近且不同种类目标跟踪情况仿真，比较仿真结果说明所提算法确实提高了多目标数据关联效果。

Research on Classification-aided Multi-target Tracking

Data association is a fundamental and challenging problem in multi-target multi-sensor tracking. When a sensor could receive classification information measurements and kinematic measurements at the same time, a classification-aided data association algorithm was proposed, which combined classification information and kinematic measurements to improve data association. The classification information likelihood function was defined by class confusion matrix, and this function was used to adjust the kenematic measurement likelihood function, so that the classification information could effectively aid the Intergrated Probabilistic Data Association. The simulation of multi-target tracking in clutter shows that the proposed technique improves data association.