Hand and elbow gesture recognition based on electromyography signal

• Main Author: Abdulhafidh Al-Dubai, Ala Abobakr
• Format: Thesis
• Language: English; English
• Published: 2019
• Subjects: TJ Mechanical engineering and machinery
• Online Access: http://eprints.uthm.edu.my/451/1/24p%20ALA%20ABOBAKR%20ABDULHAFIDH%20AL-DUBAI.pdf; http://eprints.uthm.edu.my/451/2/ALA%20ABOBAKR%20ABDULHAFIDH%20AL-DUBAI%20WATERMARK.pdf; http://eprints.uthm.edu.my/451/

This project intends to analyze and classify the Electromyography (EMG) signal of muscles that is involved in certain hand and elbow gestures. The Electromyography (EMG) data acquisition protocol is then outlined and performed where the recorded Electromyography (EMG) signal corresponds with certain hand and elbow gestures. Therefore, four hand gestures were targeted, “hand contraction, forearm rotation, wrist extension and wrist flexion”. Thus, the EMG data that have been collected from 6 subjects are compared at a small demographic scale which is age and gender. Whereas, the EMG signals are collected using the software Lab-Chart 7 with 2 channel and 5 electrodes. The pre-processing of the EMG raw signals is presented using a 6th order Butterworth band pass filter, low and high pass filter with normalization. Furthermore, the features are evaluated using Variance (VAR), Standard Deviation (SD) and Root Mean Square (RMS) to test the significance of the features. Nevertheless, the K-Nearest Neighbour (KNN) classifier is used in order to classify the EMG signals for hand gestures. Lastly, the results from this project showed that the classifier has classified the gestures with a low performance due to the fewer amounts of the subjects and some other reasons.