The Research On Data Association In Multi-target Tracking

The techniques of target tracking and data association are important topics of the research on infrared search and track system. Focus on the theory of multi-target tracking, the tracking formation, tracking maintenance, tracking termination and the design of gating are introduced, the data association are studied.Firstly, the data association of multi-target tracking is comprehensive analyzed in a point of systemic view, and a summary of the techniques of the data association appeared in both domestic and foreign publications are provided.Secondly, Some of basic data association algorithms, which include nearest neighbor data association (NNDA), probability data association (PDA), joint probability associated data (JPDA), and interactive multi-model algorithm (IMM) are in-depth studied and simulated, and the scope of its performance and adaptability are analyzed and compared.Thirdly, for the hot and difficult issues in the multi-target tracking, when the number of targets is unknown or varies overtime, this can be addressed by viewing the set of target as a random set. Based on the finite set statistics, the probability hypothesis density (PHD) filter implementation of tracking random sets is studied. Then the flow diagram of the algorithm is presented, and the performance of this algorithm is evaluated.Finally, this dissertation used probability density filter algorithm to track multiple targets in the infrared image sequence. Simulation results show that the probability hypothesis density filter can be able to estimate the number of the tracked targets in a low signal to noise ratio infrared image sequences, as well as the state of the targets.