Short-Term Hand Gesture Recognition using Electromyography in the Transient State, Support Vector Machines, and Discrete Wavelet Transform

Hand gesture recognition models determine what gesture and when this gesture was performed by a person. We propose a recognition model of hand gestures which lasted a short time (i.e., short-term gestures) using surface electromyography in the transient state, support vector machines (SVM), and discrete wavelet transform (DWT). The proposed model recognizes five gestures (wave in, wave out, fist, fingers spread, and double tap) using 8 EMG surface sensors at a sampling rate of 200Hz. Moreover, this model uses the DWT using Daubechies wavelets of order 25 as the feature extraction method, SVM with Gaussian Kernel as the classifier, and a window size of 750ms. The recognition accuracy of the proposed model is 87.5% using the EMG data of 60 people. The data analysis times of the proposed model are 382ms and 258ms using a personal computer and a server, respectively. Also, the data collection time in both devices is 750ms.