A method of target tracking

In recent years, target tracking based on Computer Vision has been used in a variety of fields. Since in most cases, motion targets are non-linear and non-gauss model, tracking algorithms based on the Kalman theory usually cannot obtain a convergent filtering result, or they cannot track the target well. However, the particle filter (PF) is an algorithm based on Monte Carlo and Bayesian theory, and it gets rid of the limitation that the model must be linear and have gauss conditions, so it can work well in these kinds of conditions. In this paper, we review the theory of target tracking, with a focus on PF, and the example of tracking human objects in an image sequence demonstrates the usage of PF.