Facerubbing: Input technique by rubbing face using optical sensors on smart eyewear for facial expression recognition

With the emergence of the wearable devices, the method to make use of the limited input space is required. This paper presents an input technique to a computer by rubbing face using optical sensors on smart eyewear. Since rubbing gesture occurs in daily life, our system enables a subtle interaction between the user and a computer. We used the smart eyewear based on the work by [5]. Although the device is developed for facial expression recognition, our method can recognize rubbing gesture independent from facial expression recognition. The embedded optical sensors measure the skin deformation caused by rubbing on the face. We detect the gestures using principal component analysis (PCA) and peak detection. we classify the area of the gesture with a random forest classifier. The accuracy of detecting rubbing gesture is 97.5%. The classification accuracy of 10 gesture area is 88.7% with user-independent training. The system can open up a new interaction method for smart glasses.