The characteristic of electroencephalograph (EEG) signal in drowsy driving was researched. Based on independent component analysis (ICA) algorithm, a method of determining the drowsiness degree was proposed. In a simulated driving system, the EEG signals of subjects, in both sober and drowsy (driving continuously for more than four hours) states, were captured by EEG instrument of NT-9200. The multi channel signals were analyzed with ICA algorithm, and removed ocular electric, myoelectric and power frequency interferences, and power spectral densities were calculated after fast fourier transform (FFT), so the fatigue index F was obtained at last. Experimental results show that the index F of drowsy state was significantly higher than the index F of sober state. The method presented in this paper can be used for determining the drowsiness degree from EEG signal effectually.
|