Person identification using components of average silhouette image

Abstract

Gait biometrics is one of the non-cooperative biometrics traits particularly in the situation of video surveillance. In the proposed method human knowledge is combined with gait information to get the better recognition performance. Here, individual contributions of different human components, namely head, arm, trunk, thigh, front-leg, back-leg and feet are numerically analyzed. The performance of the proposed method is evaluated by experimentally with CASIA dataset B and C. The effectiveness and impact of seven human gait components is analyzed by using Average Silhouette Image (ASI) under wide range of circumstances.