Study on Detection Technique for Drivers' Distraction
Zhiqiang Liu, Zhong Jing-jing · Zhongguo anquan kexue xuebao · 2010
Drivers' distraction in driving is one of the major causes of road accidents.The abnormal behaviors of drivers'head and their facial expressions were studied in detail so as to get the status information characteristics about drivers'distraction.Through making a real-time monitoring of the information on driver's facial expressions: eyes,mouth position and movement status information,the detection mechanisms was established to capture the scattered mental state of drivers and determine the driver's distracted state.Based on this achievement and with the help of the eyes and mouth region detection,BP neural network was used to estimate the different modes of driver's distraction.Meanwhile,Dempster-Shafer evidence theory was integrated to fulfill the determination of driver's distraction state by making a multi-information fusion of driver's distraction.Experimental result suggests that the technique based on BP neural network and multi-source information fusion technique with D-S rule improves the reliability and accuracy of detecting drivers' distraction state.