基于机器视觉的驾驶员注意力状态监测技术研究  

A Study on Driver Distraction Detection Technology Based on Computer Vision

作　　者：成波[1] 孟传[2] 张伟[1] 

机构地区：[1]清华大学汽车安全与节能国家重点实验室,北京100084 [2]北京科技大学自动化系,北京100083

出　　处：《汽车工程》2009年第12期1137-1140,共4页Automotive Engineering

基　　金：国家863计划项目(2006AA11Z213)资助

摘　　要：在将头部朝向近似为视线方向的前提下,研究了驾驶员头部朝向角度与注意力分散之间的关系;提出了头部朝向偏离路面的持续时间和时间比例两个注意力分散指标,建立了基于驾驶员面部特征点(如眼睛,鼻孔)的模型以进行头部朝向角度的估算和注意力分散程度的判别。实验结果表明,对驾驶员面部朝向角度的估计精度达到了较高水平,对注意力分散的检测也达到了较高的准确率。On the premise of approximating the line of sight direction by head orientation （pose）, the relationship between head pose and distraction of driver is studied. Two distraction indicators, the head-off-road glance duration and the proportion of head-off-road glance time are proposed. A model is set up based on some feature points on driver＇s face （ e. g. eyes and nostrils） to calculate the angles of head pose and to judge the extent of driver distraction. The results show that the estimation of head pose angle reaches a high precision, and the detection of distraction is also highly accurate.

关 键 词：主动安全 驾驶辅助系统 驾驶员注意力分散 机器视觉 

分 类 号：TP274.4[自动化与计算机技术—检测技术与自动化装置]