RubFinger is a novel device that can realize full-time wearable input by detecting aerial rubbing and tapping gestures of the thumb and index fingers. It uses only one small contact microphone mounted on the wrist; nothing needs to be attached on the hand or fingers. Therefore it does not interfere with daily life activities. A Hidden Markov Model (HMM) based classifier can recognize two rubbing gestures of UP and DOWN, as well as a tapping gesture of TAP with an accuracy of 97.3% under user dependent conditions. The "true one hand" operation is also realized when RubFinger is combined with wrist mounted wearables such as smartwatches.
This paper was completed by me and advised by Fukumoto san. The paper of this project was rejected by CHI2015.