Identification of Dynamic Hand Gestures with Force Myography
Eric Fujiwara, Matheus Kaue Gomes, Yu Tzu Wu, Carlos Kenichi Suzuki · 2021
Hand gestures are efficient ways to perform natural human-computer interaction. However, the current approaches rely on complex and expensive systems to recognize static poses. This work proposes a force myography sensor to identify dynamic gestures. It employs a single-channel optical fiber transducer to assess the forearm muscles, producing time-varying waveforms with distinct patterns, further processed by the classification algorithm. Assuming a set of 26 Latin letters handwritten in the air, the system provided the correct discrimination with 99.2% accuracy. Nevertheless, one may generalize this method for detecting any dynamic hand gesture, enabling applications in user interfaces, assistive technologies, and serious games.