Finger bending estimation for gesture recognition by wrist shape motion

Abstract

In recent years, the use of VR and AR has become common in various situations. Most of these controllers are still hand-held. Hand-held controllers have a problem of reducing the sense of immersion. Therefore, in this study, we focus on wrist-type controllerswhich is not hand-held controllers. Although the recognition accuracy of gestures is lower with the wrist type than with the conventional method, it is superior in terms of immersiveness and cost for gesture recognition, we propose a method to estimate the degree of finger bending by using a photo reflector to measure the wrist shape caused by finger bending, and by using the linearity obtained by comparing the finger bending with the sensor values. By using the proposed method, the finger bending can be estimated.