Flexible Non-contact Capacitive Sensing for Hand Gesture Recognition

Hand gesture recognition has become a popular research topic of human machine interface (HMI), and effective wearable sensor is an important component in the loop of hand gesture recognition system. In this paper, we introduce a flexible non-contact capacitive wristband that can be used to detect both wrist and finger gestures. To demonstrate the effectiveness and performance of the designed prototype, nine wrist gestures and ten finger gestures were selected. Five subjects participated in the experiment. To validate the importance of considering spacial relationship among channels, especially when discriminating intricate finger gestures, CNN was implemented and compared with LDA. In the wrist gesture recognition task, LDA achieved the average accuracy of 98.38%, and CNN achieved the average accuracy of 99.81%. In the finger gesture recognition task, LDA achieved the average accuracy of 90.04%, and CNN achieved the average accuracy of 95.54%. This study suggested that the designed flexible non-contact capacitive wristband could be used as an alternative for hand gesture recognition, and considering spacial relationship among channels on different measuring location yields better recognition result.