Abstract: Monitoring and analyzing the tired and attentional state by using the machine vision is one of hotspots in the assistant safe driving system.The idea of face location is introduced firstly and the techniques for face detection in this system are discussed mostly.By analyzing the actual driving state of drivers，the algorithm that based on Gray Variance locates the face of a driver，using the location of the extremums in the projective curve divides the facial organs into separate areas，deals with the areas based on OSTU algorithm and gets organs’ contours，and obtains the location of facial organs finally，is proposed and the techniques of which are introduced in detail.Practice proves that the proposed algorithm acts in real time and obtains excellent effect，which lays a good foundation for further information-collecting and state-analyzing of driver’s eyes.

Key words: machine vision, driver behavior surveillance, face organ detection, real-time

摘要： 利用机器视觉对驾驶人疲劳状态及注意力状态进行监测和分析是安全辅助驾驶领域内的研究热点之一，而对驾驶人面部器官进行准确定位则是状态分析的前提条件。首先介绍了驾驶人面部定位的思路，然后重点讨论了驾驶人面部器官检测的技术环节。通过对驾驶人实际驾驶情况的分析，提出基于灰度方差定位驾驶人面部位置，利用投影曲线极点位置分割面部器官独立区域，对每一个独立区域进行OSTU计算和轮廓提取，最终获得面部器官的位置，对算法的关键技术环节进行了详细介绍。实践证明，采用的处理算法实时性好，准确率高，效果非常理想，为下一步的驾驶人眼睛状态分析及相关信息提取打下了良好的基础。

关键词: 机器视觉, 驾驶人行为检测, 面部器官检测, 实时