摘要
随着汽车工业的不断发展,随之而来的社会问题也愈加严重。交通事故给人们造成巨大伤害的同时,也给社会带来沉重的负担和影响。由于疲劳驾驶是引起交通事故的一个主要原因。因此,研究一种合理有效、实时准确检测驾驶员疲劳驾驶的非接触式车载装置对于减少交通事故,道路安全有重大意义。 公司债券发行试点办法>为谁服务
本文研究的主要内容包括：人脸检测、人眼定位、眼睛特征提取和状态识别、疲劳程度的计算等算法的原理及实现。 首先详细阐述了经典的 AdaBoost 算法，该算法涉及的内容包括 Haar-Like 特征，弱分类器，级联的 AdaBoost 分类器等。然后利用 AdaBoost 算法进行人脸检测。梁在平
虽然 AdaBoost 算法的检测速度快，误识率低，但是在样本的权重更新过程中，对于分类错误样本中的正、负样本没有加以区分，不利于提高正样本的识别率。本文提出一种新的权重更新方法，对于分类错误的样本，对判断错误的正样本给更高的权重，使得算法在下一轮迭 代时，更加关注对分类错误的正样本的学习，从而提高对正样本（人眼）的检测率。采用基于最小二乘法对眼部的外轮廓进行椭圆拟合，根据拟合椭圆的参数来判断眼睛的睁闭状态；采用结合 PERCLOS 和眨眼频率的方法，对疲劳状态进行检测。
自动控制系统关键词：疲劳检测，AdaBoost，人眼定位，PERCLOS
Abstract
With the development of the automobile industry continuously, the social problems are more and more severe. The traffic accidents not only cause great harm to the people,but also bring heavy burden and effect to society. Because fatigue driving is a major reason that caused traffic accidents. Therefore, research a reasonable and effective real-time and non-contact device due to the detection of driver fatigue driving has great significance for reducing traffic accidents and raising road safety.
In this paper, the main research contents are face detection, eye location, eye feature extraction, principle and realization of the calculation of fatigue’ degree.
社会发展的规律
Firstly, the classical AdaBoost algorithm is produced in detail. which involves Haar-Like features, weak classifier，cascade AdaBoost classifier, etc. Next, face is detected based on AdaBoost algorithm.
Although the AdaBoost algorithm with a low false positive rate is fast, there is no distinction between samples of error classification in the procedure of weight updating,which ignores hit rate of positive samples. In this paper, a new method of weight updating is proposed to improve the hit rate of positive samples, which pays more attention to positive samples of error classification.After the region of driver’s eyes is obtained, the ellipse fitting algorithm based on least squares method is used to fit the eye contours, and the eye state is identified according to parameters of ellipse. Then, the fatigue state could be detected based on国税31号文 PERCLOS and blink frequency.
豫公网安备41160202000603
站长QQ:729038198
关于我们
投诉建议