Hand-shape classification with a wrist contour sensor: Analyses of feature types, resemblance between subjects, and data variation with pronation angle

Abstract

Hand gestures can potentially express rich information for communication between humans or between a human and a machine. However, existing hand-shape recognition methods have several problems in utilizing hand gestures in home automation. We have focused on ‘wrist contour’, and have developed a wrist-watch-type device that measures wrist contour using photo reflector arrays. In this paper, we address three challenges: improvement of hand-shape recognition performance, making clear the effect of personal difference, and identifying problems caused by pronation angle changes. To address the former two challenges, we have collected wrist contour data from 28 subjects and conducted two experiments. For the first challenge, three different feature types are compared. The results extract several important contour statistics and the classification rate is also improved by introducing multiple subjects’ data for training. For the second challenge, we compose a resemblance matrix to evaluate resemblance among subjects. The results indicate that training data selection is important to improve classification performance. To address the third challenge, two inertial measurement units are installed in the device. We have collected wrist contour data in various pronation angles, and specific relationships are found between wrist contour data and pronation angles.