Gait recognition under carrying condition: A static dynamic fusion method

When an individual carries an object, such as a briefcase, conventional gait recognition algorithms based on average silhouette/Gait Energy Image (GEI) do not always perform well as the object carried may have the potential of being mistakenly regarded as a part of the human body. To solve such a problem, in this paper, instead of directly applying GEI to represent the gait information, we propose a novel dynamic feature template for classification. Based on this extracted dynamic information and some static feature templates (i.e., head part and trunk part), we cast gait recognition on the large USF (University of South Florida) database by adopting a static/dynamic fusion strategy. For the experiments involving carrying condition covariate, significant improvements are achieved when compared with other classic algorithms.