Visual Hand Tracking Using MDSA Method

Hand tracking is a challenging problem due to the complexity of searching in a 20+ degrees of freedom (DOF) space for an optimal estimation of hand configuration. This paper represents a MDSA method (Multi-Directional search with simulated annealing) to obtain the best matching hand configurations. It uses an extended simulated annealing method (SA) with particle filter to search for optimal hand configurations in a discrete space, in which simplex search running in multi-processor (MD) is designed to predict the hand motion instead of initializing the simulated annealing randomly. The experimental results show that this method makes the hand tracking more efficient and robust.