Gait recognition based on Gabor wavelets and (2D)2PCA

Gait recognition is one of the most important techniques in application areas such as video-based surveillance, human tracking and medical systems. In this study, a novel Gabor wavelets based gait recognition algorithm is proposed, which consists of three steps. First, the gait energy image (GEI) is formed by extracting different orientation and scale information from the Gabor wavelet. Secondly, A two-dimensional principal component analysis ((2D)(2)PCA) method is employed to reduce the feature space dimension. The (2D)(2)PCA method minimizes the within-class distance and maximizes the between-class distance. Last, the multi-class support vector machine (SVM) is adopted to recognize different gaits. Experimental results performed on CASIA gait database show that the proposed gait recognition algorithm is generally robust, and provides higher recognition accuracy comparing with existing methods.