多目标跟踪是多传感器数据融合中的一个重要问题。基于模式识别理论,提出了一种通过对传感器测量数据集类,以区分源于不同目标的测量数据集合。对各个类对应的目标状态估计进行融合,从而实现了对多目标的实时跟踪。这种算法聚数据关联、数据融合和目标跟踪于一体,大大降低了多目标跟踪问题的复杂性和计算量。实际数据实验充分表明了本算法的有效性。 关键词: 目标跟踪; 数据聚类; 数据融合 Abstract : Multi2target tracking is an important problem in the field of multisensor data fusion. In this paper clustering method is applied to measurements of multisensor ,then fuse the measurements in the same cluster. A multi2target tracking algorithm has been set up by using fusion data of measurements. Practice data show that the algorithm is very effective. Key words : target tracking ; data clustering ; data fusion