The Mean Absolute Value (MAV), as the name suggests, is calculated by taking the average of the absolute value of all time samples. Since the Hodges onset of movement method rectifies the captured signal before calculating the mean, it is essentially employing MAV to determine the baseline parameter М.
Where the parameter is the number of samples and is the number of channels.
Since there are no tunable parameters, the MAV feature is limited in its ability to discriminate between classes and is usually joined with other features to make a more powerful feature vector.
The simple Square-integral method uses the energy of thr signal as a feature, calculated for signal as follows:
(4.2.)
Where is the length of the time segment following the onset of movement, . Since we are using discrete time samples, this feature can be calculated using a summation:
(4.3.)
Where is the feature, is the channel identifier, is the number of time samples, and is the sample from the th channel at time step t.