Research on hybrid tracking algorithm based on particle filter

Aiming at the problem that the traditional particle filter target tracking algorithm is poor in real time and has the degradation phenomenon, this paper proposes a hybrid tracking algorithm based on multi-technology fusion. The algorithm constructs the probability density function using unscented Kalman filter and adds the latest observations, so that the sample set is closer to the real state of the system. By employing mean shift algorithm to complete particles propagation, the particles aggregate to the high posterior probability region much further and effectively improves the sampling efficiency. The algorithm reduces the computational cost and solves the problem of degradation and real time through making the particle close to the actual state posterior distribution and increasing the number of effective particles. Simulation experiment results show that the proposed algorithm has obvious improvement in real-time and tracking error compared with the traditional particle filter tracking algorithm.