Abstract
Surface electromyogram (sEMG) has been widely used for human interface techniques. In order to make these techniques more useful and practical, it is desired that human movement is perfectly perceived and exactly duplicated from sEMG. In this study, we applied independent component analysis (ICA) to sEMG signals. ICA is a powerful procedure for a multidimensional signal analysis and has been often used for the brain research recently. Four channel electrodes were placed over near flexor digitorum superficialis and flexor pollicis longus. Subjects flexed first finger to fourth finger one by one and sEMG signals were recorded. ICA enabled extracting the independent components related to each finger flexion from these signals. Later, we were able to identify the fingers by looking at these components. Furthermore, this indicated the possibility that analysis of sEMG using both ICA and system identification may be able to identify not only the finger number but flexion force of the finger as well.
