Gait Recognition Based on Multi-Resolution Regional Shape Context

In this paper, a gait recognition algorithm based on multi-resolution regional shape context is proposed. Taking binary silhouettes as input, according to human anatomy knowledge, some key joints on human body are selected, then the shape context which counts regional pixel with multi-resolution and different support region, is used to extract the environmental characteristics of these joints, finally their mean values are computed to reflect the static and dynamic information of gait. Experiment results on NLPR database show that the proposed algorithm has an encouraging performance.