摘要
为了有效监控机动车驾驶员的疲劳状态,通过边缘检测和红色像素点提取相结合的方法确定动态阈值,对彩色人脸图像中嘴唇进行分割和提取,再利用逼近算法对嘴唇的各个特征点进行定位,采用机器视觉监视嘴唇运动,对驾驶员的疲劳状态进行判定。试验结果表明:该定位算法可有效去除嘴唇周围的干扰,完整地将嘴唇从原图中提取出来,并准确进行嘴唇特征点定位;该算法对肤色深浅不同、光照不均匀、头部偏转角度不同等情况均有很好的鲁棒性。
In order to monitor driver fatigue state effectively,its dynamic threshold was determined by using edge-detection method and red color element point extraction method,lip was segmented and extracted in colorful face images.Each feature point of lip was located based on approximation algorithm.The movement of lip was monitored,and the state of driver fatigue was determined based on machine vision.Test result indicates that the locating algorithm can dislodge the interferences around lip effectively,pick up lip shape from original image completely, and locate the feature points of lip exactly. The algorithm has good robustness for the situations of different skin colors, different lights and different rotational angles of head. 10 figs, 9 refs.
出处
《交通运输工程学报》
EI
CSCD
北大核心
2009年第4期118-121,126,共5页
Journal of Traffic and Transportation Engineering
基金
国家自然科学基金项目(10402008)
关键词
交通信息工程
驾驶员疲劳监控
嘴唇特征点定位
机器视觉
边缘检测
traffic information engineering
driver fatigue monitor
locating of lip feature point
machine vision
edge detection
