A fast and highly accurate EMG signal classification approach for multifunctional prosthetic fingers control

Nowadays, modern prostheses, which is defined as myoelectrical controllers, are used that can be controlled by using signals from muscles instead of traditional prostheses. In this paper, the electromyography (EMG) signals recorded when ten finger movements including one finger, two fingers and hand close movements were performed are classified using features by statistical methods. EMG signals were obtained from 10 participants who were sitting constantly with their arms supported. According to the results, the classification accuracy was achieved an average 81.60% by using K-Nearest Neighbor, average 98.94% by using Linear Discriminant Analysis (LDA). As a result of the analysis, it has been detected that LDA, which is suggested method, has classified the most accurately.