Facial and bio-signal fusion based driver alertness system using Dynamic Bayesian Network

Car accidents are a major threat to drivers which happens because of fatigue and we have no effective alertness system to detect them, previous methods have concentrated only on facial recognition that includes eye closure, blink-frequency to determine the number of times he is closing his eyes, eigenfaces modeling which takes multiple images of a person under different light illumination condition. These methods have not concentrated towards the biological condition of the person which can normally be detected by ElectroCardioGraph(ECG), ElectroEncephaloGraphy (EEG). So this paper concentrated towards face recognition by using Support Vector Machine (SVM) algorithm and bio-signal by calculating saturation of oxygen (Sp02) in blood determined by a Pulseoximeter and also the pressure applied on steering determined by pressure sensor, defines a threshold value for determining fatigue and to improve performance we fused the facial and bio-signal values using Dynamic Bayesian Network which is an effective algorithm to detect fatigue. A warning system sounds if driver fatigue is believed to reach a defined threshold. We can also consider the pressure given by the driver on the steering and on applying brakes which will vary according to driver´s biological condition.