A Fast Gait Recognition Method Based On Feature Curve Matching

Gait recognition means to recognize the identity of a person according to his walking posture. As an emerging biometric identification method, gait recognition has attracted more and more attentions by its merits of identifying a people at the condition that far distance, noncontact, low resolution. And it’s hard to conceal and disguise. In this paper, a fast and efficient method of gait recognition for a special gait database is designed, and the main contributions are as follows:1. An algorithm of measuring the similarity between the features of two gaits based on the feature curve is proposed. Several major joints in a body are chosen as the feature points of a gait, so the location of each feature point in the entire image sequence can be connected to a feature curve, then ICP(Iterative Closest Point) algorithm is used to calculate the matching degree between each two corresponding feature curves of two different gaits. Besides, a method of calculating the weight of every matching degree is proposed to compute the similarity of two different gaits through weighting sum every curve matching result. Finally, the gait similarity algorithm was evaluated by using leave-one-out method on the database, and achieving the satisfactory results.2. In this paper, several features of a gait are fused to the identification. The relative height and stride of a person on the image are used to search the people whose relative height and stride meet the conditions, after searching, the identity range is much smaller. Next, the gait similarity algorithm is used to determine the identity of the person by choosing the highest similarity. Last, each gait sequence of the database was carried out identification test, moreover, the identification process of several gait sequence were carefully monitored. The experimental results reveal, the gait recognition method has a fast recognition speed and high recognition accuracy.3. The software design of the gait recognition system has been completed, including several functions of the database operation, drawing feature curves, the evaluation of gait similarity algorithm, identity recognition and so on, all this greatly promote the gait recognition system into practical application.