Abstract
Hand motion tracking and gesture recognition are of crucial interest to the development of virtual reality systems and controllers. In this paper, a wireless data glove that can accurately sense hands’ dynamic movements and gestures of different modes was proposed. This data glove was custom-built, consisting of flex and inertial sensors, and a microcontroller with multi-channel ADC (analog to digital converter). For the classification algorithm, a hierarchical gesture system using Naïve Bayes Classifier was built. This low training time recognition algorithm allows categorization of all input signals, such as clicking, pointing, dragging, rotating and switching functions when performing computer control. This glove provided a more intuitive way to operate with human-computer interface. Some preliminary experimental results were presented in this paper. The data glove was also operated as a controller in a First-Person Shooter (FPS) game to perform the usability of the proposed glove.