Signal whitening preprocessing for improved classification accuracies in myoelectric control

The surface electromyogram (EMG) signal collected from multiple channels has frequently been investigated for use in controlling upper-limb prostheses. One common control method is EMG-based motion classification. Time and frequency features derived from the EMG have been investigated. We propose the use of EMG signal whitening as a preprocessing step in EMG-based motion classification. Whitening decorrelates the EMG signal, and has been shown to be advantageous in other EMG applications. In a ten-subject study of up to 11 motion classes and ten electrode channels, we found that whitening improved classification accuracy by approximately 5% when small window length durations (<; 100 ms) were considered.