摘要
为了减少由于驾驶员疲劳驾驶引起的交通事故,提出驾驶员疲劳状态检测系统的方案。使用3×3中值滤波去除噪声和光照对图像的影响,通过对AdaBoost算法的强分类器训练算法改进、级联分类器优化实现人脸的快速检测,在检测到的人脸区域,通过积分灰度投影和从粗到细改进的模板匹配方法对人眼进行准确定位;通过PERCLOS、眼睛闭合时间、眼睛眨眼频率、嘴巴张开程度、头部运动的计算,进行驾驶员疲劳程度的综合判定。实验结果表明,该方法准确率高,兼具了良好的实时性和鲁棒性。
In order to reduce the traffic accidents caused by driver fatigue, this paper proposes a method of the driver fatigue detection system. This method uses median filtering to remove the impact of image noise and light, then achieves rapid detection of human faces by the improved strong classifier training algorithm of AdaBoost algorithm and the optimized cascade. In the detected face region, it uses gray projection points and an improved from coarse to fine template matching method to implement the positioning of human eyes accurately. Using calculated PERCLOS, closure eye time, eye blink frequency, degree of mouth opening and the movement of head, it can determine the comprehensive degree of driver fatigue. The experimental results show that this method has high accuracy with a good real-time and robustness.
出处
《计算机工程与应用》
CSCD
2013年第15期253-258,共6页
Computer Engineering and Applications
关键词
疲劳检查
人脸检测
人眼定位
fatigue detection
face detection
eyes location
