Velocity-robust gait recognition based on stochastic switched auto-regressive model

Gait recognition is a promising non-intrusive biometric method. By using kinematical cues, a new gait recognition model which synthesizes dynamic model and statistical model is proposed. The proposed model allows us to establish security applications without the aids of shape cues deriving from marker-dependent motion capture cameras, for it only exploits low-dimensional kinematical cues. Angular variables of ankle joint are adopted as the model´s input, and a 2-link virtual passive walking model plays an important role both in the configuration of the parameter matrix and the selection of the matrix´s initial values. By evaluation the recognition rates of different models, the velocity-robust characteristics of the new model and its low computational load compared with conventional HMM are verified. Furthermore, the dynamics of human walking, the kinematical cues derived from the dynamics model, and any accessible position cues extracted either from pictorial data or ranging sensor can be integrated into a coherent system.