摘要
为了解决现有的疲劳状态检测算法在光线不足时准确率较低的问题,提出了一种在光照不足环境下疲劳驾驶检测的算法。首先对采集到的人脸图像进行低照度图像增强处理,以此来提升图像的曝光度;然后采用YOLO人脸目标检测算法获取到人脸的区域;之后采用人脸特征点检测算法勾画出脸部特征点进行睁闭眼和打哈欠的判断;最后统计单位时间内眼睛闭上的比值与上打哈欠的次数,以此判断驾驶员是否在疲劳状态。实验结果表明,在较暗的环境中,提出的检测算法相较于传统算法取得了更好的效果。
In order to solve the problem that the existing fatigue state detection algorithm has low accu-racy rate when the light is insufficient,an algorithm for fatigue driving detection under insufficient light environment is proposed.Firstly,the acquired face images are processed with low-illumination image enhancement as a way to improve the exposure of the images;Then the YOLO face target detec-tion algorithm is used to obtain the area of the face;After that,the face feature point detection algo-rithm is used to outline the face feature points for the judgment of open and closed eyes and yawning;Finally,the ratio of closed eyes per unit time plus the ratio of yawning is counted as a way to determine whether the driver is in fatigue state.The experimental results show that the proposed detection algo-rithm achieves better results compared with the traditional algorithm in a darker environment.
作者
詹林
张跃
刘唤唤
ZHAN Lin;ZHANG Yue;LIU Huanhuan(School of Computer Science and Engineering,Anhui University of Technology,Huainan 232001,Anhui,China)
出处
《合肥学院学报(综合版)》
2023年第2期90-94,共5页
Journal of Hefei University:Comprehensive ED
基金
安徽省自然科学基金“传输矩阵-相位匹配法研究金属纳米颗粒阵列/光学薄膜光学特性”(1808085QF205)。
关键词
疲劳驾驶
人脸检测
低照度图像
特征点
YOLO
fatigue driving
face detection
low illumination images
feature points
YOLO