Individual authentication through hand posture recognition using Multi-Hilbert Scanning Distance

Abstract

In this paper, we propose a novel Hand Posture Recognition (HPR) for biometrics. This study uses the three dimensional point clouds for robust hand posture recognition at the rotation and scale. Multi-Hilbert Scanning Distance (MHSD) are also introduced for mathematical approaches of shape matching. HPR framework is divided into five parts: detecting hand region, removing the wrist, aligning the hand pose, extracting feature descriptor, and matching. Based on the experimental results, this framework showed superior results for hand posture recognition rate.