Target tracking using the snake particle filter

This paper presents a method, the snake particle filter (SPF), for tracking targets in video sequences. Manual or semi-automated solutions are both expensive and susceptible to error. In the SPF algorithm, automated tracking is accomplished by combining the particle filter with the snake. Here we employ the snake to establish the target shape, which is used to assign the weight for each particle in the particle filter. The snake provides a likelihood measure in the flexible particle filter framework that accommodates non-linear, non-Gaussian systems. Our results show that the SPF algorithm has an associated low RMSE value of approximately five pixels in the sequences tested for this study.