基于眼态识别的疲劳驾驶检测技术研究-交通信息工程及控制专业论文.docxVIP

I I 摘 要 随着人们生活水平的提高、生活节奏的加快，由于疲劳驾驶导致的交通事故也在逐 年增多。驾驶员处于疲劳状态时对车辆的操作控制能力以及对周围环境的感知判断能力 都会有所下降，因此非常容易引发交通事故。疲劳驾驶已经成为当今交通安全的重要隐 患之一，因此研制一套实用性强、检测准确率高的疲劳驾驶预警系统具有重要的现实意 义。 本文根据疲劳驾驶预警系统的技术特点和应用要求，设计出了一套疲劳检测系统的 改进方案，该设计方案主要包括以下几个处理过程：第一步，采用摄像头获取驾驶员上 身视频图像；第二步，对检测图像进行初始图像处理，包括光照补偿、背景差分等，然 后利用 Adaboost 算法对驾驶员人眼部位进行精确定位；第三步，采用 Mean Shift 算法 与 Kalman 滤波器相结合的方法来进行人眼窗口的跟踪以及位置预测；第四步，应用椭 圆拟合法来进行人眼轮廓的提取以及应用霍夫变换来实现人眼虹膜的定位，并应用这些 检测结果来判断驾驶员眼睛的闭合情况以及注视方向；第五步，应用 PERCLOS 结合人 眼注视方向这一眼态参数的方法来实现疲劳状态的识别。 基于以上算法流程，本文设计了一套在 PC 环境下的疲劳驾驶仿真预警系统，根据 实验结果，本疲劳驾驶仿真实验系统能够实时检测出人眼闭合情况以及视线注视分布情 况，能够较准确的判断出被检测者是否处于疲劳驾驶状态。 关键词：疲劳驾驶，图像处理，Adaboost，Mean Shift，Kalman，霍夫变换，PERCLOS II II Abstract With the improvement of peoples living standard and the speeding up of peoples life rhythm, traffic accidents due to fatigue driving also increased year by year. When driver is in the state of fatigue, his control ability to the operation of the vehicle and the perceptual ability on the surrounding environment will fall. So its very easy to cause traffic accidents. Fatigue driving has become one of the most important hidden troubles of the traffic safety today. Therefore developing a set of practical, high detection accuracy of fatigue driving warning system has important practical significance. In this paper, we designed a set of improved solution of the fatigue driving testing system according to the technical feature and the application requirements of the fatigue driving early-warning system. The design mainly includes the following process: Step one, get video images of the upper part of drivers body using cameras. Step two, do the initial image processing for the detection images, including illumination compensation, background subtraction, and so on. Then locate the driver eyes accurately using the Adaboost algorithm. Step three, track eye picture and predict the location of it using Mean Shift algorithm combined with Kalman filter. Step four, Extract the human eye contour using ellipse fitting method and locate th