An enhanced multi-view human action recognition system for virtual training simulator

Virtual military training systems have received considerable attention as a possible substitute for conventional real military training. In our previous work, human action recognition system using multiple Kinects (HARS-MK) has been implemented as a prototype of virtual military training simulator. However, the classification accuracy of HARS-MK is not enough to be utilized for virtual military training simulator. In addition, the experiments are carried out under just two simple action types; walking and crouching walking. In order to overcome these limitations, in this paper, we propose an enhanced multi-view human action recognition system (EM-HARS). Compared to HARS-MK, in EM-HARS, feature extractor is enhanced by employing covariance descriptor. In addition, the feasibility test of EM-HARS is conducted under various human actions including military training actions which are newly captured. The experiment results show that EM-HARS achieves higher classification accuracy than that of HARS-MK.