Fractal dimension for drowsiness detection in brainwaves

Drowsiness, the state where a person´s alertness is reduced, is one of the major causes of accidents. Therefore, there is a need for the detection of this state of human brain, especially for the people working in industries and also for those involved in driving activities. Hence, drowsiness detection system is gaining more importance these days. In this paper, the authors have presented a drowsiness detection methodology using electroencephalogram (EEG) signals. The method used in the detection is fractal dimension (FD), which is a measure of irregularity of the curve. Three FD algorithms, Higuchi, Katz and Petrosian are tried along with logarithm of energy (log E), and Intrinsic dimension (ID) on a set of 15 alert and 15 drowsy signals. In addition, certain statistical features are extracted from the signals. The Support vector Machine (SVM) classifier used in this work yielded a sensitivity of 76 percentage, a specificity of 70 percentage, in distinguishing the drowsy and alert samples.