基于SMC-CPHD的多目标跟踪算法研究

针对CPHD滤波算法在多目标跟踪中计算难处理和对于局部目标估计存在漏检的问题，提出了基于序贯蒙特卡罗方法的基数概率假设密度（SMC-CPHD）滤波算法。这种方法是将SMC和CPHD两种滤波算法的优点相结合，用一些离散的粒子去接近PHD函数，不仅解决了在滤波修正步没有闭式解的问题，而且避免了当某个目标发生漏检时，PHD权值的转移问题，在递推PHD函数的同时也递推基数分布。将此方法应用到有杂波存在复杂的多目标跟踪环境中，通过仿真实验，对CPHD滤波和SMC-CPHD滤波得出的结果进行比较，验证了本文所提出方法对多目标跟踪的可行性和精确性。

Multi-Target Tracking Algorithm Based on SMC-CPHD Filter

In order to solve problems of computational intractability and local missed detection in multiple targets tracking of Cardinalized probability hypothesis dentisy(CPHD) filer,a(CPHD) filter algorithm based on Sequential monte carlo(SMC)is proposed.The advantages of SMC and CPHD filter are combined in this method to make some discrete particles close to the PHD function,which not only solve the problem of lack of closed-form solution in filter corrector,but also avoid the weight transfer problem of PHD filter when some targets miss detection,so that the cardinality distribution can be propagated while the PHD function is propagated.This algorithm is applied to the complex multitarget tracking environment under clutter and simulated.The simulation results based on comparison between CPHD filter and SMC-CPHD filter verify the practicability and accuracy of the proposed method in multitarget tracking.