Abstract:

This paper proposes a simple but efficient algorithm of gait recognition. In each image sequence, background subtraction and related processes are car  ried out to segment and track the moving silhouettes of a walking figure from the background. By computing the angles of the pixels in the moving silhou ettes, we can get the angle histogram vectors as the gait signature. Supervised pattern classifiers with Euclidian Measurement are finally performed for recognition. Extensive experinacnts using the NLPR database demonstrate that the proposed algorithm has a high recognition performance with low computa tional cost.

Key words: biometrics, gait recognition, background subtraction, angle histogram