Head Pose Tracking with Individual Adjustment for Communication Tools

In this paper, we propose a novel method to automatically estimate diffusion parameters within particle filter for efficient head state tracking. Recently, many head pose and facial expression tracking methods have been developed. However, most of the conventional methods are not designed as communication tools. We therefore propose a new head state tracking method which can be applied to various applications such as avatar's facial expression control. The proposed method optimizes the diffusion parameters found within particle filter which is utilized for tracking via the maximum-likelihood approach. Moreover, we propose an update number function that is utilized for incrementally updating the data set. Although the diffusion parameters vary according to the head motion, accurate tracking with the suitable parameters can be realized.