Research On MeMber Extended Target Tracking Based On Box Particle Filter

Target tracking technology has a very wide range of applications, especially in the military field such as cruise missiles, intercontinental missiles, precision strike, and fire control of aircrafts and ships. With the progress in science and technology, the resolutions of modern radars and sensors have been greatly improved, large or close targets generate a number of measurement data, which causes extended target tracking problem. Extended target tracking needs the partition of targets' measurements and appropriate filtering algorithms.In this dissertation, on the basis of interval analysis theory, multiple extended target tracking using the box particle filter is researched, which focuses on the Cardinality Balanced Multi-target Multi-Bernoulli(CBMeMBer) Filter for extended target tracking.1. The characteristics and methods of extended target tracking are studied. The characteristics, measurement partitioning and commonly applying partitioning methods of the extended target are presented.2. The box particle filter(BPF) is researched. The box particle filter is a combination of interval analysis theory and Monte Carlo algorithm. It replaces the point particles with box particles whose maximum errors are known. It can effectively treat imprecise measurements with error range. One advantage of the box particle filter is that it can reach the same filtering precision with dozens of box particles whereas the particle filter needs thousands of particles, thus can greatly reduce the amount of computation and improve the operation efficiency.3. The particle filter based CBMeMBer filter needs a large number of particles in order to obtain accurate tracking performance, and computations reduce the efficiency of the algorithm. In this dissertation, an extended target CBMeMBer algorithm based on box particle filter is presented. Simulation results show that the new algorithm gives more accurate estimate of the number of targets, decreasing the amount of computation compared with the particle filter based CBMeMBer algorithm, thus is more suitable for engineering applications.4. When the existence probability of missed targets is large, the innovation information is weakened in the CBMeMBer algorithm. An extended target tracking IMeMBer algorithm based on box particle filter is presented. By adding a decision step, IMeMBer algorithm can solve this problem. The principle and procedure of BPF particle filter based IMe MBer algorithm is presented. Simulation results show that the proposed algorithm has the advantages of a shorter operation time and reliable results.