Gait recognition using Gait Entropy Image

Gait as a behavioural biometric is concerned with how people walk. However, most existing gait representations capture both motion and appearance information. They are thus sensitive to changes in various covariate conditions such as carrying and clothing. In this paper, a novel gait representation termed as Gait Entropy Image (GEnI) is proposed. Based on computing entropy, a GEnI encodes in a single image the randomness of pixel values in the silhouette images over a complete gait cycle. It thus captures mostly motion information and is robust to covariate condition changes that affect appearance. Extensive experiments on the USF HumanID dataset, CASIA dataset and the SOTON dataset have been carried out to demonstrate that the proposed gait representation outperforms existing methods, especially when there are significant appearance changes. Our experiments also show clear advantage of GEnI over the alternatives without the assumption on cooperative subjects, i.e. both the gallery and the probe sets consist of a mixture of gait sequences under different and unknown covariate conditions.