Gait recognition using Gabor-based enhanced active energy image

Feature extraction is a challenging task in gait recognition, because of the dynamic process of the human gait. Active Energy Image (AEI) is proven as an effective gait representation. For extracting discrininate dynamic information from AEI, an enhanced AEI is proposed for gait representation. By using the enhanced parameters Y, the more discriminate information is obtained. To address the problem of the appearance changes (such as clothes, carrying bags) of human gait, the EAEI is then represented in the low subspace by Gabor-based discriminative common vector analysis. To evaluate the proposed approach, some experiment on CASIA dataset B is designed. The results prove its effectiveness in terms of recognition rate. © 2009 by Binary Information Press.