A Multimodal Vigilance Monitoring System Based On Fuzzy Logic Architecture

This paper deals with the problem of vigilance level monitoring. A novel method of hypovigilance detection is presented in this work. It is based on the analysis of eyes' blinking and head posture. The fusion task of both systems is achieved by the fuzzy logic technique which allows us to obtain five vigilance levels. This paper contains two key contributions. The first is the amelioration of our previous works in the classification field employing fast wavelet network classifier (FWT) by using another classification system based on a deep learning architecture. It provides more accurate results than the wavelet network classifier. The second resides in the conception of a driver alertness control system able to detect five vigilance levels which is different from previous works of the literature characterized by two, three or four levels. Experiments, using different datasets, prove the good performance of our new approach.