Optimal filtering for noise reduction in myoelectric signals on an embedded system

Electromyography (EMG) signals are used to diagnose muscular pathologies and are also employed as inputs for electronic applications. However, a major disadvantage in detecting this signal is the noise that can be derived from diverse sources such as power supplies, incorrect placement of the electrodes, and the environment. To reduce the noise in the signal a filter must be incorporated in the system. The aim of this paper is to reduce the noise generated in the muscular signals through an embedded system, using an optimal filter configured with different frequencies and numbers of samples values. A real-time application was implemented using an Olimex SHIELD-EKG-EMG shield and a SAM V71 board. The shield was used to obtain the signal using the ADC of the SAM V71 board and the filter was programmed on the board. A correct configuration of the frequency and number of samples is crucial to compute an optimal filter on an embedded application. The noise reduction results are presented.