【摘要】 为了避免日常生活中因疲劳造成的交通事故,利用人眼图像识别技术对驾驶进行疲劳监测。设计并提出了一套基于图像信息判断驾驶员疲劳状态的流程。首先使用Viola-Jones算法,在图像中直接进行人眼定位,再对人眼图像进行处理,得到人眼长宽比值Q,最后建立了Q值和疲劳状态的关联模型。实验证明,该定位方式比传统方式检测速度提高了50%以上,同时能适应头部不同姿态;建立的"眼部相对长宽比-睁眼程度"模型具有良好的线性。模拟实际驾驶环境进行测试结果表明,监测系统能适应不同受试者并在1 s内对疲劳状态的驾驶员发出警报,在防疲劳安全驾驶领域有一定的应用前景。更多还原

【Abstract】 In order to avoid accident caused by fatigue,the system is designed to monitor driver’s fatigue based on human eye image recognition technology. A process to determine the degree of fatigue with image information is designed. First directly locate human eye via Viola-Jones algorithm,then get width-to-length ratio Q of human eyes,and finally established a correlation model of Q and fatigue. As experiment shows,the proposed eyelocating method with capability of different head postures runs more than 50% much faster than conventional ways; the established "Q-eye open level"model has good linearity with adaptation to different drivers. Real driving environment is simulated for testing,the system can give alarms when driver’s fatigue is detected within 1second,which bestows a place in the field of anti-fatigue safe driving applications.更多还原