【作者】 张利伟；

【导师】 张航；

【作者基本信息】 中南大学 ， 控制科学与工程， 2009， 硕士

【摘要】 由于疲劳驾驶导致的交通事故数量不断上升,疲劳驾驶已成为一个严重的社会问题。因此,研制疲劳驾驶监测预警系统,对于避免交通事故,提高交通系统的安全性有着重要意义。本文针对基于计算机视觉的疲劳检测方法进行研究,构建了一个实时的驾驶疲劳检测系统。该系统可以通过CCD摄像头实时地采集驾驶员上半身的视频图像到计算机,然后通过计算机视觉算法捕获驾驶员的眼睛闭合度和嘴唇特征来估计驾驶员的疲劳程度,检测出疲劳。本文研究的核心内容包括:人脸定位、眼睛定位、眼睛跟踪、眼睛信息提取、嘴唇特征提取和疲劳度计算等算法的实现与改进。在人脸检测方面,提出了肤色检测与面部特征空间位置验证相结合的方法。首先针对人脸肤色在色彩空间中的分布及其特性展开研究,建立了有效的肤色模型,对图像进行肤色分割;接着,对分割结果依据人脸的矩特征去掉部分非法的肤色区域;最后,在所得区域内检测出面部特征,按照特征之间的空间位置关系对候选脸区的真假进行验证。经实验表明,该方法在保证较高的检测率的基础上,大大降低了误检率。在眼睛定位方面,采用了基于Haar-Like特征级联分类器的检测方法,在人脸区域内按照不同尺度搜索存在的眼睛。在眼睛跟踪中,提出了一种基于卡尔曼滤波与Mean Shift算法相结合的眼睛跟踪方法,此法能够进行尺度更新和准确地跟踪闭眼。在眼睛信息提取方面,在分析了相关的研究方法基础上,本文提出一种计算眼睛闭合程度的新方法—虹膜外接矩形法,其原理是通过一个外接矩形绑定虹膜区域,然后用该矩形的宽高比值来估计眼睛闭合度。与以往的算法相比该方法可以更精确地表示眼睛的状态。在疲劳度计算方面,提出综合眼睛信息和嘴唇特征计算疲劳程度的方法,并在PC机上对算法进行了仿真实验,取得了良好的实验结果。更多还原

【Abstract】 The increasing number of traffic accidents due to a driver’s fatigue has become a scrious problem for the society.With the background, developing a system for monitoring the driver’s level of vigilance and alerting the driver when he fatigues is essential.This paper chooses computer vision based method as the study target to build a real-time detection system of driver fatigue.The system has a CCD camera which collecting videos of driver’s upper-body to a computer,and uses some algorithm of computer vision to compute the percentage of eyelid closure and lip features,and then,estimates the level of driver fatigue.The core content of this paper is that realizing and improving algorithm such as face location,eye location,eye information extraction ,lips information extraction and fatigue level calculation.In order to detect frontal face,a method which combined skin color detection and validated spatial facial features scope is presented.The distribution and characteristic of skin color is firstly studied in color space to establish effective skin color model and to use it in image segmentation.Then,some unseemliness regions in segmentation are eliminated according to moment features.At last,we locate the facial features based on the acquired skin color,and to validate whether this face region contains face or not according to the relative location among features.The experiment results show that this method decrease false detection rates enormously.In order to locate eyes on the face, a Cascade Classifier based on Haar-Like features is used to search all existing eyes in the face region in different scale. Fourthly, Kalman filtering and mean shift algorithm based eye tracking method which can track close eyes and change size of eyes is proposed.In order to get eye information extraction,some relative researches have been analyzed,before a novel method,named as iris’s external rectangle based method,is proposed to calculate the percentage of eyelid closure.The principle of the iris’s external rectangle based algorithm is that: bind the iris region by using an external rectangle,use the ratio of the width and height of the rectangle to estimate the percentage of eyelid closure.This method can represents eye states more precisely than others relative method.In order to compute percentage of eyelid closure, eye information and lip features is used to evaluate the fatigue level of driver.The results of tests based on PC show good ability of this research.更多还原