Frontal motion automatic tracking based on scaled prismatic model
Tao Hong, ShenKang Wang · 2004
Automatic model initialization and motion tracking is one of the most important parts of human motion analysis in monocular image sequences. In this paper, a frontal motion tracking method was presented based on the combination of the standard particle filter algorithm and a model initialization technique. The scaled prismatic model was employed and the state parameters in the particle filter were initialized several times using the model initialization technique during the frontal motion tracking. Experimental results show that the proposed method outperforms the standard particle filter algorithm since it can recover from tracking failures and reduce computational load.