An Effective Driver Fatigue Monitoring System

This paper proposes an effective driver fatigue monitoring system. A Haar-like based cascaded AdaBoost classifier is trained for eye region localization from the input drive face video; and then lattice degree of nearness based on Fourier descriptor is utilized for eye states identification; finally, PERCLOS is calculated for fatigue detection. The most prominent contribution of this paper is: instead of localizing each eye accurately, some useful contour and edge features are extracted by DFT, and lattice degree of nearness is introduced to determine eye states without difficult threshold problems in many traditional eye states algorithms. The algorithms presented in this paper are proved to be both robust and fast for driver monitoring system by a large amount of experiments.