An eigen based feature on time-frequency representation of EMG

In this research we used a multi-channel electromyogram acquisition system using programmable system on chip (PSOC) microcontroller from previous work to acquire surface EMG signals. The two channel surface electrodes were used to measure and record EMG signals on forearm muscles. These two channels of EMG signals were performed a blind signal separation by using an independent component analysis (ICA) technique. The well known ICA algorithm called FASTICA is a useful method to separate two or more linear combination of source signals into statistically independent components. We purposed A novel features for the EMG contraction classification. Our feature is derived from the application of time-frequency analysis of the EMG signal followed by the computation of Eigen vector of the timefrequency magnitude spectrum. Our feature is the ratio between the two Eigen values. We have shown the robustness of our features for a variety of muscular contraction. The result is very promising. © 2009 IEEE.