The attempt to recognize patterns from the EEG potentials is at the base of the field of the Brain Computer Interface. 

Processing EEG signals mainly involves two steps:

1)   Relevant EEG features must be extracted from raw signals. 
2)   These features must be effectively transformed (classified) so that the output control signal        (command) is as close  as possible to the user’s will.  

Motor imagery is defined as mental rehearsal of a motor act without any overt movement execution. The recognition of the neuroelectric or hemodynamic patterns underlying motor imagery has been demonstrated to be one of the fundamental steps in developing brain–computer interface (BCI). 

Typically, a BCI is a communication system able to discriminate different neuroelectric or hemodynamic patterns in real time (Wolpaw et al., 2002). In this regard, several evidences have been collected on people who can learn to control electroencephalographic (EEG) mu or beta rhythm amplitude generated over their sensorimotor areas, by using mental imagery of motor actions to control physical or virtual devices in one or two dimension (Schlogl et al., 2005).