基于神经网络的驾驶员觉醒水平双目标监测法

在驾驶员疲劳状态单目标状态识别的基础上,提出了驾驶员嘴巴与眼睛状态双目标疲劳状态判别法.该方法将驾驶员眼睛和嘴巴的特征向量值按先后顺序,分别输入到修正的BP网络中,采用区域匹配算法,根据驾驶员疲劳状态量化评判标准,综合识别驾驶员的觉醒水平.采用VC++开发了驾驶员疲劳监测算法软件,并对驾驶员进行了监测仿真试验.结果表明,双目标监测系统能够实时、准确地监测和识别驾驶员的觉醒水平,具有较高的识别容错性和准确性.

Double-Objective Detection Based on Neural Network for Driver's Alert Level

Based on the single-objective detection,a double-objective detection is proposed for driver's alert level,i.e.,the states and changes of both eyes and mouth of a driver are taken as the drowsy discriminance during monitoring a driver's fatigue through his/her entire facial image when traveling.In this way both the characteristic vectors of eyes and lip of a driver are sequentially input into a modified BP network.Then,the region matching algorithm is introduced to detect comprehensively the alert level of a driver in accordance to the quantified evaluation indices showing how he/she is feeling fatigued.VC is used to develop an algorithm software to monitor driver's fatigue,with which a simulation test is carried out.Test results show that the proposed double-objective detection system can monitor and identify accurately a driver's alert level with high fault tolerance provided.