A contactless hand shape identification system

Geometric measurements of human hand shape have been used for personal authentication in a number of commercial systems. Yet, traditional hand shape image acquisition devices have auxiliary pegs or planar to help users place their hands correctly for accurate identification. It is not so acceptable for privacy reasons. Moreover, the contact may contaminate and corrode the device. Aiming at these disadvantages, a contactless hand shape identification approach is proposed in this paper. Hand images are captured by the self-developed device in a contactless manner. After that, geometry features are extracted based on key points, which are selected by Sliding Window Filtering Algorithm. Distance measuring is then performed on feature vectors to get a final decision. Experiments show that Equal Error Rate of the system reaches 2.16% and 2.40% in two different kinds of data sets.