Techniques Of Multiple Passive Sensors Multiple Targets Tracking Based On Random Finite Sets Theory

The techniques of multiple targets tracking based on multiple passive sensors are the important topics of the research on passive target tracking system, which have been found with wide applications in both military and civil areas. But because of the improvements of flight performances and electronic countermeasure (EC), tracking environment has changed significantly. The passive multiple targets tracking are faced with great challenges of strong target maneuvering, high clutter, low detection probability, high false-alarm ratio and an unknown time varying number of targets etc.To solve the problems discussed above, this dissertation mainly involves researches about the multiple passive sensors multiple targets tracking algorithms based on Random Finite Sets (RFS) statistics. Firstly, the methods of data processing in multiple passive sensors system are investigated. For a fixed number of non-linear multiple targets tracking, a novel particle filter based on fuzzy-probability weighting algorithm in multiple passive sensors system is proposed which can achieve higher tracking precision. Secondly, the RFS statistics also the methods of Probability Hypothesis Density (PHD) of RFS is implemented in multiple targets tracking. For tracking an unknown time varying number of multiple targets passively in heavy clutters environment, algorithms of Gaussian Particle filtering PHD (GSPPHD) are studied. In order to improve the tracking performance, an improved GSPPHD algorithm is proposed. Simulation results illustrate that the improved algorithm has a higher tracking precision and lower computational complexity. Furthermore, For tracking an unknown time varying number of multiple maneuvering targets in multiple passive sensors system, an interacting multiple model (IMM) implementation of the PHD algorithm is proposed. Simulation results are presented to show the effectiveness of the proposed filter over some single-model PHD filters. Finally, due to the deficiency of estimation for multiple targets tracks with the PHD filters, a novel Rao-Blackwellized particle filter (RBPF) used as an association and filtering algorithm based on RFS theory is proposed. By combination with PHD filter, both the multiple targets tracking and the targets tracks association are achieved effectively.