Improved Joint Probabilistic Data Association Method based on Interacting Multiple Model

Abstract

Multitarget tracking with highly maneuver and crossing track under dense-clutter environment is an emphasis in target tracking field. To effectively differentiate the measurement of cluster and target, and establish the relation between target and measurement, the data relation technologies are needed to keep the consistency of path tracking. When maneuvers happen to the target, a suitable motion model should be selected for self-adaptive tracking. With consideration of these two points, we combine JPDA algorithm with IMM algorithm according to some certain way, to track multiple maneuvering targets in clutter environment. Then an improved method named OEA-JPDA (Once Echo Association) is proposed. OEA-JPDA selects the model corresponding to the biggest determinant of the covariance matrix as the target moving model of current time. Simultaneously, a related cluster matrix is created. When the amount of echoes and models is large, all the filters are performed echo associated once with added prediction consolidation and probability update. The simulation results show that this method has higher tracking accuracy for multiple maneuvering targets which are changed with time, under the dense-clutter scenario. Even if strong maneuver occurs, there is no need for bigger adjustment to the mean square error for target destinations. The improved scheme can achieve effective tracking in cluster environment with lower computation