Study Of Multi-Target Particle Filter Algorithm

The detection and tracking of weak targets is one of the great challenges for modern radar systems. A new signal processing technology——tracking before detect (TBD) in which the particle filter TBD technology emerges, as the current weak target detection and tracking for the better performance of a processing algorithms, it has become a hot TBD technology to be researched. As a technology being developed, especially multi-target particle filter TBD algorithm, has many problems to be resolved, such as the interference of closely spaced targets and engineering applications on multi-target particle filter TBD algorithm.Regarding to the problems mentioned above, the multi-target particle filter TBD tracking algorithm which based on TBD technologies and the engineering application of multi-target particle filter algorithm are studied. The main work is as follows:1. Analyzed the theoretical of the particle filter, researched the single-target particle filter TBD algorithm, which makes the theoretical foundation for the study of multi-target particle filter TBD algorithm.2. Aiming at the problem of the performance reduces heavily when targets is nearby on multi-target particle filter TBD algorithm, proposed point spread function independent partition algorithm, mitigated the performance degradation when targets is nearby, verify the improved one has better performance under the same conditions through simulation experiments.3. Aiming at the engineering problems of particle filter multi-target TBD technology, analyzed the computation and storage of the multi-target particle filter TBD algorithm, given relationship among the number of particles, target number, algorithm performance, hardware resource requirements. Additionally two kinds of hardware implement schemes are proposed, which are serial operation using single processer and paralell operation using multipal processers, and then analyzed the advantages and disadvantages of both hardware implementation and the conditions applicable to them,which laid the foundation for multi-target particle filter TBD algorithm works to achieve.4. Aiming at the limitations what classic multi-target particle filter TBD algorithm does not apply to the scene that the number of targets is unknown and the changes over time, used r structure, achieved IP and PSFIP two-layer particle filter when the number of targets is unknown and the changes over time, and made a comparison, verified PSFIP two-laye algorithm can estimate the number of targets more accurate when there are targetst nearby.The multi-target particle filter TBD algorithms proposed above have been verified its feasibility.by simulation experiment and measured data.The analysis of the hardware resource requirements and the two hardware implementations to be proposed on the algorithm is ready for the realization of multi-target particle filter TBD algorithm.