A Study for Gender Classification Based on Gait via Incorporating Spatial and Temporal Feature Matrix

Gait recognition is a new biometric identification technology. In this paper, we present a study and analysis in incorporating spatial and temporal feature matrix of gender classification based on human gait, which can increase the robustness of temporal gait feature. Gait period is obtained by locally linear embedding (LLE) algorithm. A different way based on Principal Component Analysis (PCA) to get gait feature, which is called Gait Principal Component Image (GPCI), was used. And GPCI is a grey image. K Nearest Neighbor (KNN) classifier was used for gender classification. The experimental results show that the proposed approach achieves a high gender classification accuracy rate and prove that spatial and temporal feature fusion is rational and valid.