Gait Recognition Considering Directions of Walking

Studies on gait recognition are mostly based on the assumption that walking direction is parallel to the camera, and the person's side view is extracted. Walking direction has becoming one of the gait recognition challenge problems. In this paper we explore gait recognition considering any directions of walking which is not definitely parallel to the camera. We propose a novel approach to computing the walking direction and extracting features by employing a human model. Furthermore, a support vector machine (SVM) is performed allowing us to investigate and evaluate the recognition power of any walking directions. We applied our method to the real human walking video data, and achieved high recognition rate. Our approach finds out how changes in walking direction affect gait parameters in terms of recognition performance. As it is entirely based on human gait, our approach is robust to different type of clothes and different walking directions