Fusion of multiple gait features for human identification

Gait recognition has recently attracted increasing interest from the biometric community. In this paper, we propose a simple new feature called multi-bipolarized contour mean (MBCM) for gait recognition. The proposed MBCM feature consists of four components: (1) the vertical positive contour mean, (2) the vertical negative contour mean, (3) the horizontal positive contour mean, and (4) the horizontal negative contour mean. We fuse the proposed gait features at a feature level to improve recognition performance. The proposed recognition system is evaluated with the NLPR gait database.