[1]梁元辉,吴清乐,曹立佳.基于多特征融合的眼睛状态检测算法研究[J].计算机技术与发展,2021,31(02):97-100.[doi:10. 3969 / j. issn. 1673-629X. 2021. 02. 018]
　LIANG Yuan-hui,WU Qing-le,CAO Li-jia.Research on Eye State Detection Algorithm Based on Multi-feature Fusion[J].,2021,31(02):97-100.[doi:10. 3969 / j. issn. 1673-629X. 2021. 02. 018]
点击复制

基于多特征融合的眼睛状态检测算法研究()
分享到：

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
31
期数:
2021年02期
页码:
97-100
栏目:
图形与图像
出版日期:
2021-02-10

文章信息/Info

Title:
Research on Eye State Detection Algorithm Based on Multi-feature Fusion
文章编号:
1673-629X(2021)02-0097-04
作者:
梁元辉1; 2; 吴清乐1; 2; 曹立佳1; 2
1. 四川轻化工大学 自动化与信息工程学院,四川 宜宾 644005;?
2. 四川轻化工大学 人工智能四川省重点实验室,四川 宜宾 644005
Author(s):
LIANG Yuan-hui1; 2; WU Qing-le1; 2; CAO Li-jia1; 2
1. School of Automation and Information Engineering,Sichuan University of Science & Engineering,Yibin 644005,China;?
2. Key Laboratory of Artificial Intelligence of Sichuan,Sichuan University of Science & Engineering,Yibin 644005,China
关键词:
眼睛状态监测; 疲劳驾驶; 多特征融合; PERCLOS; EAR
Keywords:
eye state detection; drowsing driving; multi-feature fusion; PERCLOS; EAR
分类号:
TP391
DOI:
10. 3969 / j. issn. 1673-629X. 2021. 02. 018
摘要:
疲劳驾驶检测算法研究对提升交通安全有着重要的意义。目前,已有大量关于疲劳驾驶的文献和成果。在疲劳驾驶检测算法中,眼睛开闭状态的判断起着至关重要的作用。深度级联卷积神经网络用来检测人脸和人脸特征,利用 Dlib 工具快速提取驾驶员人脸特征。基于眼睛特征计算眼睛宽高比,并将眼睛宽高比、传统人眼特征的人眼虹膜等用于判断眼睛开闭的参数。论文提出一种实时融合了 EAR、虹膜等多个特征的眼睛状态检测算法,可补偿传统人眼特征的像素值比较敏感的不足, 也补偿了 EAR 在人脸倾斜、戴眼镜、光照变换、眼睛周围有光斑等情况下非常不可靠的不足。 在640*480 分辨率,帧率 30 fps 的视频上获得平均 92% 的检测正确率。 实验结果表明融合后的算法可在光照变换、人脸倾斜、佩戴眼镜等条件下提升检测性能,鲁棒性较高。
Abstract:
The research about driving drowsiness detection algorithm is of great significance to improve traffic safety. Presently,there are many literatures and achievements about driving drowsiness. In driving drowsiness detection algorithm,the judgment of eye state plays animportant role. A deep cascaded convolutional neural network to detect faces and face features,and Dlib tool to quickly extract drivers’face features. Eye aspect ratio (EAR) and pupil are used to detect eye stature. We propose a real-time eye state detection algorithm thatintegrates EAR,pupil and other features,which can compensate for the lack of relatively sensitive pixel value of tradi-tional human eyefeatures and compensate for the unreliability of EAR in face tilt,glasses wearing,light transformation,light spots around the eyes andother situations. The average detection accuracy is 92% in 640*480 resolution and 30 fps video. The experi-ment shows that the proposed algorithm can improve the detection accuracy especially in light transformation,face tilt,glasses wearing,etc.with high robust ness.
更新日期/Last Update: 2020-02-10