嵌入式疲劳驾驶监测系统的设计与实现

疲劳驾驶是引发恶性交通事故的重要原因之一,驾驶员疲劳监测技术近年来已逐步成为图像处理领域的一个研究热点。基于改进的和提出的新算法,设计了一个嵌入式驾驶员疲劳监测系统。由可见光/近红外摄像头采集视频,首先采用Haar特征的级联分类器从图像中检测出人脸区域,并用钻石搜索法跟踪人脸区域;然后提取一个新的图像差分统计特征,并结合3个准则判断疲劳状态;最后采用全变分模型消除图像中的非均匀光照,以便实现鲁棒的人眼定位和人脸识别。实验测试结果表明,本系统的疲劳状态监测准确率达到95%以上。

Design and Implementation of an Embedded System for Driver Fatigue Surveillance

Fatigue driving is one of important reasons that cause fatal traffic accidents.Technology for driver fatigue surveillance is becoming a new research topic in image processing and application fields.An ARM embedded system is designed for driver fatigue surveillance based on some improved and proposed algorithms.Video is captured by a camera with the frequency range comprising of visible light and near infrared ray.From the video,face region is firstly detected by the cascaded classifiers based Haar features and traced by a diamond searching algorithm,eye state is then judged by three criterions based on a novel statistical image difference feature, and uneven lighting is compensate by a total variation model for robust eye location and face recognization.Experimental results show that the correct rate for driver fatigue surveillance is beyond 95% in our embedded system.