Wavelet moments for gait recognition represented by motion templates

This paper reports a novel research that uses the wavelet invariant moments of the motion templates representing walking figure to determine representative features for gait recognition. Changes of pedestrian silhouettes covering one stride are decomposed into disappear and emergent image area, from which motion history is cumulated into static grey scale images as motion templates. These motion templates are employed to characterize human walking properties. In order to reduce the feature space, the cubic B-spline wavelet moments are computed from the motion templates and extract discriminating features for identification. Experimental results show that the proposed approach is an effective and efficient gait representation for individual recognition, and the proposed approach achieves highly competitive performance with respect to other approaches.