Pattern recognition of finger movement detection using Support Vector Machine

Abstract

This paper describes signal processing of surface electromyography (sEMG) for finger movement detection. Stoke survivors could use this application to retrain or helping them in their activities. This assistive technology will help them in order to improve the functional capabilities. The signal processing in this experiment is using 256Hz sampled data of sEMG signal. Three fingers of right hand is detected by using three channels of sEMG signal sources. System using Butterworth bandpass filter to eliminate noises. The filter using cut-off frequency 10Hz and 40Hz. Some features for the detection is built from statistical approach. System is using Support Vector Machine (SVM) to detect and classify fingers movement by using those features. From experiment, the accuracy of the sytem is about 98.3%.