Infrared target tracking base on auxiliary particle filtering algorithm

In order to solve the problems of complex calculation in the infrared target tracking, the auxiliary particle filtering algorithm was built by the utilization of Bayesian importance sampling algorithm, the introduction of auxiliary particle variables on the basis of large weight particles, and the redefinition of importance sampling distribution function to prevent the change of the particle probability density after re-sampling. The two-weighted calculation makes the change of the particle weight ratio more stable and the sampling point closest to the true state only by the particle weight obtained from the resampling, in which the probability threshold of particles at different weight values can be taken as the criterion for judging whether the particle filtering has been completed. In the infrared moving target model structured in the two-dimensional plane, the system is zero-mean Gaussian white noise. Simulation data shows that the algorithm is superior to the particle filtering and re-sampling particle filtering algorithms in the mean square error in x and y directions, picture processing, RMSE performance.