Gabor filter bank-based GEI features for human Gait recognition

This paper proposes a supervised feature extraction approach which is capable to select distinctive features for the recognition of human gait under clothing and carrying conditions thus improving the recognition performances. The principle of the suggested approach is based on the use of feature texture descriptors extracted from Gait Energy Image (GEI). The proposed features are computed using the bank of Gabor filters and then selected using Spectral Regression Kernel Discriminant Analysis (SRKDA) reduction algorithm. The proposed method is evaluated on CASIA Gait database (dataset B) under variations of clothing and carrying conditions for different viewing angles; and the experimental results using one-against-all SVM classifier have given attractive results of up to 91% in terms of Correct Classification Rate (CCR) when compared to existing and similar state-of-the-art methods.