Cepstral based features for gait recognition

This paper presents a novel approach of gait identification system based on cepstral analysis. For each gait sequence, Radon Energy Image (REI) and Gait Energy Image (GEI) are obtained. Cepstral analysis is applied to the energy images to extract the de-correlated discriminant gait features. The experiments with a multi-view point environment and shape variation condition show an improved identification results. The proposed technique is evaluated using 105 people in CASIA-B database in five view angles. Experiments based on the score level fusion of different view angles achieved a correct recognition rate of 99.04% using cepstral GEI.