Hierarchical approach for articulated 3D human motion tracking using PF-based PSO

In this paper, particle filter integrates with particle swarm optimization (PF-PSO) is proposed for articulated 3D human motion tracking. In vision-based human motion tracking, two algorithms most extensively have been used, namely, PF and PSO. In order to take the advantage of both algorithms, we use the PSO algorithm in the particle filtering to shift the weighted particles toward into high probable space to get the optimal human pose. In order to reduce the computational cost we optimize the body poses in hierarchical manners. The approach shows strength in the qualitative comparisons with other conventional state-of-the-art algorithms like PF, annealed particle filter, and PSO. © 2014 WIT Press.