Abstract: A JPDA based algorithm is presented for real-time tracking of dim moving point targets in image sequences according to the geometric relationships between confirmed measurements and multi-target validation gates.The validation matrixes are used to calculate all joint events and the corresponding parameters，regardless of whether the measurements fall inside intersection of several validation gates.The joint association probabilities are calculated between each measurements and their possible source targets according to the JPDA algorithm，and then the joint association probabilities are combined with the Kalman filter to accomplish the task of predicting and updating the status of each target.Theoretical and experimental results show that this algorithm has high real-time tracking performance，and provides high tracking accuracy.

Key words: Joint Probabilistic Data Association（JPDA）, dim point target, validation matrix, joint events, association probabilities, real-time tracking

摘要： 根据确认的众多量测和众多目标跟踪窗之间的几何关系，引入确认矩阵并计算所有联合事件及其对应的参数，不论量测是否落入跟踪窗相交区域，根据JPDA算法计算每一个量测与其可能的各个源目标之间互联的概率。将互联的概率与Kalman滤波器相结合从而完成对每一个目标的预测和更新。理论及实验结果表明，该算法适用于序列图像密集杂波环境下的全程跟踪，并取得了一定的理论和仿真结果。

关键词: 联合概率数据关联（JPDA）, 点状目标, 确认矩阵, 联合事件, 关联概率, 实时跟踪