Research, Long-range Identification Algorithm Based On Human Gait

Gait recognition is the new field of the Biometrics Recognition. Early psychological studies into gait suggested that gait was a unique research involves in following areas: movement segmentation, feature extraction, pattern classification and gait database and so on. In this paper, the force three areas are mainly researched.The movement segmentation is the foundation of the gait recognition. On the gait contours segmentation algorithm based on a new spatio-temporal combination, the two value result which was gained by combining background subtraction and symmetric frame difference performed moving estimation, it was the new method of time-domain segmentation; the first order wavelet transform was used to the current frame image, watershed segmentation algorithm was used to divide the LL weight image which was extended into many closed and non-overlapping regions, it was space-domain segmentation; at last, maked the result of space-domain segmentation casting shadow to the result of time-domain segmentation. On the extraction approach to gait contour based on improved C-V model, still used the new method of time-domain segmentation to extract the mostly entire contours of human-body, the contours serve as the primary zero level set of improved C-V model to get the aim of gaining the accurate contours of human-body by a handful of the frequency of the iteration.The kernel of the gait recognition is the gait feature, GEI further reflected the static information of the human-body, and the periods template of the gait feature of the Zernike moments further reflected the moving information of the human-body, effectively combined the two kinds of the gait feature to get aim of the commutative complementarity.The pattern recognition is the last step of the gait recognition, the feature data gained by the GEI and the period template of the gait feature of the Zernike moments was reduced by the PCA, and gain the performance of gait recognition. In the final, used the fusion of multiple-characteristics in the decision-making layer to improve the performance of gait recognition.The proposed algorithm in this paper was experimented in the NLPR gait data, the performance of gait recognition achieved more than ninety-five percent. The result of the experiment proved that the methods proposed in this paper has stated practicality value.