Abstract
In recent years, the more types of computers appeared, the more situation that users input have changed the variety. Sometime in the future, we should acquire input method, free from any devices or location. We propose a new recognizer for hand gesture recognition using a wearable device. Finger is connected with each muscle in each of the tendons, so it can be moved by muscle contraction and extension. We use Myo, wearable device, for measuring the surface electromyograph (SEMG) that would have occurred from muscles corresponding to each of the tendons. To get the movement information of the finger, we use the LeapMotion at the same time. It is carried out independent component analysis (ICA) using the measured SEMG and fingers' motion, to create a cross-section map of the forearm. With the map for estimating finger and/or hand status from SEMG, hand gesture recognition rate will be improved.
