Automated hand shape verification using HMM

In this paper we present a method for identity verification based on matching of hand shapes. Our method first represents the shapes of hands by sets of ordered points. Then the contour of the hand is characterized by a features sequence consisting of two parameters: the radius and curvature at the contour points, MMM has proved a very successful tool for modeling and recognition sequence signal. So the hand shapes are compared using HMM. We apply a normalization score measurement to improve the classification ability and robustness. The experiment results show the effectiveness of our method and the correct verification rate can be above 90%.