CLASSIFICATION OF HAND GESTURE BASED ON EMG SIGNALS

Abstract

This study implemented classifiers for hand movement classification based on EMG signals from the forearm muscles to create a command interface for a myoelectric hand prosthesis. The classifiers used were Support Vector Machine (SVM), K-Nearest Neighbours (KNN), and Artificial Neural Networks (ANN). The classifiers trained for each participant and the group as a whole. Principal Component Analysis (PCA) is used to improve classification results. All classifiers achieved better performance in individual cases. PCA did not contribute to the improvement of classification. SVM achieved the highest accuracy values for most participants compared to the other classifiers. Sensitivity results indicated that SVM and KNN can recognize wrist extension movements for all participants. After PCA, ANN showed low deviations among participants in classifying wrist extension movements.