【作者】 王军；

【导师】 李景宏；

【作者基本信息】 东北大学 ， 电路与系统， 2009， 硕士

【摘要】 随着生活水平的提高,交通运输事业的发展,道路交通事故逐年增多。频繁发生的交通事故中,疲劳驾驶是引发交通事故的主要原因之一,其造成的交通事故占交通事故总数的10%以上。因此,如何准确、快速的判断驾驶员的疲劳状态,对减少交通事故的发生具有重大意义。本文重点研究了人脸检测、人眼定位以及在其基础上的疲劳判断。疲劳驾驶检测系统主要由分类器训练模块、图像采集模块、图像处理模块、目标检测模块和疲劳判断模块五部分组成。利用Adaboost算法完成分类器训练模块,可以在目标图像中检测出人脸,定位出人眼的位置。图像采集模块是通过摄像头来采集目标图像的视频帧,获得原始的目标图像。图像处理模块对摄像头采集到的目标原始图像进行灰度处理、图像二值化、图像增强和图像去噪声处理,得到待检测的目标图像,以便更加准确的检测出人眼位置。分类器训练模块训练的分类器加载到系统中在目标检测模块中完成对目标的检测。疲劳判断模块是利用PERCLOS算法,完成对定位的眼睛进行疲劳判断。通过自适应算法不断地改进阈值的设定,以提高疲劳检测系统的性能。采用了Microsoft Visual Studio 2005开发环境来开发疲劳驾驶检测系统的识别和检测算法,实现时用到了Intel的OpenCV库,在研祥智能的EC5-1719CLDNA嵌入式开发板上实现,并对系统进行了相关的测试。整个系统基本实现了对目标的疲劳状态检测的功能,能对正面、侧面和戴眼镜的目标进行判断,具有较高的准确性。同时,系统不仅应用到驾驶员的疲劳检测,也可以对一些监控室、大型控制室等场合的操控人员进行疲劳检测,来减少因疲劳而导致的重大损失。更多还原

【Abstract】 With the improvement of living standards and the development of transport, the road traffic accident is increasing year by year. Fatigue driving is one of the major causes in the frequent occurrence of traffic accidents and accounted for more than 10% of the total of the traffic accidents. Therefore, how to determine the status of the driver’s fatigue more accurately and quickly is great significant to reduce the occurrence of traffic accidents.This article focuses on the human face detection, the eye location, as well as in the judgments of fatigue on the basis of them. Fatigue driving detection system mainly consists of five parts, including the classifier training module, image acquisition module, image processing module, target detection module and fatigue determined module. Adaboost algorithm is used to complete the classifier training module, the human face can be detected in the target image, and the location of the human eye can be positioned by the algorithm. The image acquisition module is used to collecte the original image of the target by the camera head. The image processing module is for the gray-scale processing to the original image from the camera, the binarization of the image, the enhancement of the image and the processing to remove noise from the image. So we can get the target image which is to be tested, and it can detect the location of the human eye much more accurately. The classifier of the training module is loaded into the system, and it is to complete the target detection in module. By the PERCLOS algorithm, we can determine the fatigue eyes which are positioned, and that is the fatigue determining module. In addition, in order to improve the performance of fatigue testing system, we do continuously improvement of the setting of threshold based on the adaptive algorithm.Microsoft Visual Studio 2005 development environment is used to develop fatigue detection system of identification and detection algorithms, the OpenCV the Intel libraries is used to achieve the system, and the system-related test is also done. The whole system basically achieves the function of testing fatigue state on the target, and it has the capabilities of detecting the targets which is positive, the side and wearing glasses with high accuracy.At the same time, the system is not only applied to the driver’s fatigue detection, but also can test the fatigued from the manipulated staff who are in the Monitoring rooms and some large-scale controlling rooms and so on. Therefore, the system can reduce heavy losses caused by fatigue.更多还原