Feature Extraction Through Wavelet De-Noising of Surface EMG Signals for the Purpose of Mouse Click Emulation

An electromyographic-based method for detecting user "mouse clicks" is presented which is designed to minimize user energy expenditure and operational false positive rate. Current EMG-based methods use basic threshold crossing and do not consider user fatigue, nor sensor and biometric noise. In the present method, background EMG noise, sensor noise, and eye-blinks are removed from raw data such that any feature remaining can be served as a "click" event. This implementation is shown to be effective at removing sensor-based noise, background EMG and adapting to user fatigue. We demonstrate true "mouse click" events are detected against eye-blinks event when they are similar in amplitude