An ARM-based embedded gesture recognition system using a data glove

Gesture is a natural and intuitive mode of interpersonal communication, which can simulate language and behaviors to express certain meanings and words. In this paper, an embedded gesture recognition system has been designed and developed using a data glove. The design of a data glove is aimed at supplying a auxiliary communication tool for deaf people so that there is more social intercourse between them and the normal. The gesture recognition system which can interpret gesture to voice and text messages is mainly composed of four parts: a data glove, an ARM processor, a display module and an audio module. The data glove comprises five unidirectional bend sensors (FLX-03) and a 3-axis accelerometer. The ARM processor receives the data that is transmitted from the sensors of a data glove through I/O ports. And it also analyzes and calculations the data from distinct gestures, then compares the data with the templates to see if the gestures match. The recognition results are displayed on an LCD screen, and the converted voice is outputted through an external speaker. Combined with the hardware, this paper focuses on the template matching algorithm, including data collection, mining and processing. The experimental results show that the recognition accuracy rate of trained gestures is above 91%.