基于计算机视觉的驾驶员疲劳状态检测预警技术研究

疲劳驾驶是造成交通事故的主要原因之一,为提高驾驶员疲劳驾驶状态的智能化检测水平,提出一种基于计算机视觉的面部多特征疲劳驾驶检测算法。该算法采用多线程优化后的Dlib（图像处理开源库）实现对驾驶员面部的定位与追踪,利用Dlib开源库中的人脸关键点检测器对驾驶员面部关键特征点进行提取,实时计算驾驶员眼部的纵横比（EAR）和嘴部长宽比（MAR）,并以自制视频流数据集作为实验样本计算出相关阈值,有效提高了检测算法的普适性,在此基础上,计算出眨眼频率、闭眼次数、眼睛闭合时间百分比（perclos）以及打哈欠频率这四个反映驾驶员疲劳状态的指标,并利用数学方法进行指标实时融合,根据融合指标的数值对驾驶员疲劳状态进行分级,最终通过实验验证该疲劳检测系统的准确性。结果表明,提出的综合疲劳指标能够准确反映在不同环境和光照下驾驶员的疲劳状态和发展趋势,驾驶员疲劳判定的正确率达到97.5%以上。

Research on Driver Fatigue State Detection and Early Warning Technology Based on Computer Vision

Fatigue driving is one of the main causes of traffic accidents. In order to improve the intelligent detection level of drivers" fatigue driving state, a facial multi-feature fatigue driving detection algorithm based on computer vision was proposed. The algorithm uses DLIB (open source library of image processing) optimized by multi-thread to realize the location and tracking of drivers" faces. The face key point detector in the open source library of DLIB is used to extract the key feature points of drivers" faces. The aspect ratio (EAR) and aspect ratio (MAR) of drivers" eyes are calculated in real time. And using the self-made video streaming data set as the experimental samples to calculate the relative threshold, effectively improve the universality of the detection algorithm, on this basis, calculate the percentage of blinking, number of close your eyes, eyes closed time (perclos) and yawning frequency these four indicators reflect the drivers" fatigue state, and by using mathematical method to index real-time fusion, According to the value of the fusion index, the fatigue state of the driver is classified. Finally, the accuracy of the fatigue detection system is verified by experiments. The results show that the proposed comprehensive fatigue index can accurately reflect the fatigue state and development trend of drivers under different environments and illumination, and the correct rate of driver fatigue determination reaches more than 97.5%.