Biometric identification using an electroencephalogram signals compression based on KC function

الملخص EN

EEG (electroencephalogram)is brain waves measure.

the brain troubles are evaluated by EEG.

it is used to locate of the activity in the brain during a seizure and to consider the patients whose suffer from brain functions problems.

these troubles include tumors, coma, confusion and long-term difficulties (such as weakness associated with a stroke).

the acquisition of EEG signals requires contact and liveliness and these signals are changes under stress that make so potentially unnecessary if it is acquired under menace.

in this paper, an innovative and robust solution for this problem referred to earlier was introduced.

to get this goal, the manner depends on models of various data compression models of information-theoretic plus the metrics symmetry related to kolmogorov complexity.

the proposed procedure make a compare between two EEG segments the procedure able to cluster the data in three collections: corresponding record, different participant, self-participant for the stratification of the proposed measure with values close to 0 for the same participant and closer to 1 for the different participants (entrants).

the strategy was carried out to determine the participant in the database based on EEG signals.

A 1-NN classifier was implemented, using a distance measurement method proposed in this scheme.

the classifier was able to correctly identify almost all participants, with 96% accuracy in the underlying database.