Gait Recognition Based on GFHI and Combined Hidden Markov Model

Gait as a biometric that can be collected at a long distance. This feature has many potential applications in monitoring fields. In this paper, the gait representation method of the gait optical flow history image (GFHI) is proposed by combining the optical flow of gray and the gait history image, which realizes the overall and compact localized representation of human motion. We proposed a gait recognition method based on a combined hidden Markov model (HMM), using two groups of hidden Markov models to distinguish similar gait sequences. The experimental results show that the method in this paper has a high recognition accuracy. In a small gait dataset, the average recognition accuracy can reach 0.74, 0.94, and 0.96 when the training data is single, dual, and three-view.