基于眼动与脉搏信息融合的驾驶疲劳识别算法研究

发布时间：2018-03-31 10:38

本文选题：信息融合　切入点：驾驶疲劳　出处：《山东大学》2015年硕士论文

【摘要】：交通安全是与国计民生直接相关的热点问题,其中驾驶疲劳作为导致交通事故多发的主要问题之一,受到了人们的广泛关注。驾驶疲劳会导致驾驶员精力不集中、肢体反应迟钝、对突发情况的应变能力下降,极易引发交通事故。据统计,由驾驶疲劳所造成的交通事故约占事故总数的20%,占重特大交通事故的40%,占交通死亡率的80%。由此可见,驾驶员疲劳已经成为交通事故的主要隐患,而且不容易被察觉和监测。为此,大量学者对驾驶疲劳问题展开了研究,并取得了一定的研究成果。为了解决驾驶疲劳识别过程中所存在的信号源单一、识别精度低、实用性能差等不足,本文对基于眼动与脉搏信息相融合的驾驶疲劳识别算法进行了研究。该算法以眼动和脉搏两种信息作为研究对象,分别对两种信息进行疲劳特征提取,并以极限学习机作为分类器,对驾驶员是否处于驾驶疲劳状态进行识别。本文的主要工作包括：1)对驾驶疲劳的形成原因、社会危害、研究现状,以及当前研究中所存在的问题进行了分析和总结。2)对眼动信息进行处理,深入分析了混合积分投影和mean-shift两种算法的原理,设计了基于混合积分投影与mean-shift相结合的眼部疲劳状态特征提取方法。该方法包括背景去除、人眼定位、人眼分割和特征计算四个步骤,能够综合考虑人眼的灰度信息和位置信息,准确地提取出PERCLOS、眨眼频率和平均闭合时间三个眼动疲劳特征。3)对脉搏特征进行处理,通过类比于心电的疲劳特征,选择主波间期均值、主波间期标准差、高低频功率比三个参数作为脉搏疲劳特征,对脉搏信号与疲劳状态之间的联系进行了研究,证明了脉搏信号可以应用于驾驶疲劳的识别,且具有较好的识别效果。4)设计了基于极限学习机的驾驶疲劳识别算法,通过极限学习机对所提取的眼动和脉搏特征进行融合,并对驾驶员是否处于疲劳状态进行识别。通过设计实验,采集被试者在模拟驾驶环境中的眼动和脉搏数据,对算法的实际性能进行验证。实验结果表明,该算法具有良好的识别率。
[Abstract]:Traffic safety is a hot issue directly related to the national economy and the people's livelihood, among which driving fatigue, as one of the main problems that lead to frequent traffic accidents, has attracted widespread attention. Driving fatigue can lead to driver's inconcentration. The body is slow to react, and its ability to respond to emergencies decreases, so it is easy to cause traffic accidents. According to statistics, traffic accidents caused by driving fatigue account for about 20 of the total number of accidents, 40 percent of the serious traffic accidents, and 80 percent of the traffic mortality. Driver fatigue has become a major hidden danger in traffic accidents, and it is not easy to detect and monitor. Therefore, a large number of scholars have carried out research on driving fatigue. In order to solve the problems of single signal source, low recognition accuracy and poor practical performance in the process of driving fatigue identification, some research results have been obtained. In this paper, a driving fatigue identification algorithm based on the fusion of eye movement and pulse information is studied. And using the extreme learning machine as the classifier to identify whether the driver is in the state of driving fatigue. The main work of this paper includes the causes of the formation of driving fatigue, the social harm, and the current research situation. And the existing problems in the current research are analyzed and summarized. 2) the theory of hybrid integral projection and mean-shift algorithm is analyzed. An eye fatigue feature extraction method based on mixed integral projection and mean-shift is designed, which includes four steps: background removal, human eye location, human eye segmentation and feature calculation. Considering the gray level information and position information of human eyes, we can accurately extract three eye movement fatigue features: PerCLOS, blink frequency and mean closure time. 3) to process the pulse characteristics, and compare to the fatigue characteristics of ECG. The main wave interval mean, the main wave interval standard deviation and the ratio of high and low frequency power are selected as the pulse fatigue characteristics. The relationship between the pulse signal and the fatigue state is studied. It is proved that the pulse signal can be applied to the identification of driving fatigue. And has a better recognition effect. 4) designed the driving fatigue identification algorithm based on the extreme learning machine, through the extreme learning machine to extract the eye movement and the pulse characteristic to carry on the fusion, Through the design of experiments, the data of eye movement and pulse of the subjects in simulated driving environment are collected, and the actual performance of the algorithm is verified. The experimental results show that, The algorithm has good recognition rate.
【学位授予单位】：山东大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U495;U463.6;TP202

【相似文献】