HumTouch: Finger gesture recognition on hydrogel-painted paper using hum-driven signals

Abstract

We propose a passive touch sensor that can be implemented on semi-conductive materials, such as paper. This sensor detects electric currents leaked from a human body when touched with bare fingers. Such electric currents are mostly produced by AC power lines in the environment. We tested the potential of the sensor for gesture recognition by using a paper sheet painted with a hydrogel ink. A gesture classification algorithm combining the $k$ -means method and Mahalanobis' generalized distance could successfully classify 90% of the gestures drawn on the paper with a fingertip. Gestures including multi-directional lines and circles were nearly perfectly classified whereas digits, such as 2 and 8 exhibited difficulties with classification. The hydrogel ink can turn other non-conductive materials comprising woods and plastic into touch sensors, and such hum-based touch sensors are expected to find practical applications in gesture recognition.