Abstract
This paper proposes a motion capturing system for human walking in the side view. First we build a 3D human model with structural and kinematical constraints. The model is build by OpenGL and viewed as the candidate model. To track the human motion parameters, we use the separated particle filter for tracking six parts of human body. This method can obviously reduce the high-dimensional parameters. Second we use the Particle Filter (PF) and Nonparametric Belief Propagation (NBP) for human tracking. PF will estimate some initial pose, and then NBP will compute the results after several iterations. Then the results will be viewed as the initial value for the next stage of particle filter. Finally, we can compute the motion parameter of each frame. Then error angle of our system is less than 11 degrees.