基于人脸识别的驾驶疲劳检测方法研究-生物医学工程专业论文.docx

河北工业大学硕士学位论文 河北工业大学硕士学位论文 基于人脸识别的驾驶疲劳检测方法研究 基于人脸识别的驾驶疲劳检测方法研究 基于人脸识别的驾驶疲劳的检测方法研究 摘 要 随着经济的快速发展，车辆的不断增多，交通事故有不断上升的趋势，给人们的生命 财产安全带来了极大的危害。驾驶疲劳是导致交通事故的重大因素之一，因而对驾驶疲劳 的检测和控制不仅具有重要的学术意义，而且具有重要的社会价值。 本论文首先介绍了驾驶疲劳的检测方法，指出基于人脸识别的驾驶疲劳检测的可行性 和可靠性。接着设计了模拟驾驶实验，采集 3 位模拟驾驶人员的脸部正面图像作为实验图 像数据。进而，对模拟驾驶实验中得到的人脸图像数据进行处理和分析，并对驾驶员的疲 劳状态进行检测。具体地，首先利用 AdaBoost 算法检测驾驶员的人脸图像，并利用模板 匹配的算法定位驾驶员的眼睛；然后提取眼睛特征向量；最后把眼睛的特征向量送入支持 向量机中进行分类。实验结果显示，利用 AdaBoost 和模板匹配算法的结合对眼睛的识别 率达到 90%以上，利用支持向量机中对眼睛状态的分类准确率达到 91%以上。这表明本文 的方法能够很好地对驾驶员状态进行疲劳判断。本论文完成了对驾驶疲劳图像检测实验方 案的设计，基本实现了基于人脸识别的驾驶疲劳的检测。 关键词：驾驶疲劳，人脸识别，AdaBoost算法，模板匹配，支持向量机 i PAGE PAGE iv RESEARCH ON DRIVING FATIGUE DETECTION METHODS BASED ON FACE RECOGNITION ABSTACT With the rapid development of economy and the rising number of vehicle, the rising trend of traffic accidents has brought great harm to peoples life and property safety. Driving fatigue is one of the major factors of traffic accidents. Thus the detection and control of traffic accidents caused by driving fatigue is not only valuable in academy, but also important in social sense. In this thesis, driving fatigue detection methods are introduced firstly, and the feasibility and reliability of the driving fatigue detection based on face recognition are pointed out. Then the front face images of three simulation drivers are collect as the experimental image data in simulated driving environment. Furthermore, the algorithms are designed for the processing and analysis of the image data in the simulations, and the fatigue states of the drivers are detected. Concretely, the drivers faces are detected using AdaBoost algorithm first and the drivers eyes are located using template matching algorithm. Then the eye feature vector is extracted, and we put it into the support vector machine for classification. The experimental results show that the eyes’ discrimination using the combination of AdaBoost and template matching algorithm reaches around 90%. The classification of the eyes states reaches over 9