Driver Fatigue Detection Based on Unscented Kalman Filter and Eye Tracking
-
摘要: 为解决驾驶员疲劳检测算法中头部快速移动、人眼非线性跟踪以及实际疲劳表情的识别问题,提出了一种新的基于UKF眼跟踪算法的驾驶员疲劳检测方法.根据近似非线性函数的概率分布比近似其函数更容易的原则,利用UT无迹变换,选择一组确定的Sigma点集逼近驾驶员人眼运动状态的后验概率密度函数,进行人眼非线性跟踪.在驾驶员人眼非线性跟踪基础上,通过计算PERCLOS值,进行现实驾驶条件下驾驶员疲劳的跟踪检测.实验结果表明,该方法不仅可以增强对驾驶员头部旋转、快速移动以及光照变换的鲁棒性,而且可以比传统的Kalm an滤波算法提供更精确的计算估计.Abstract: In order to resolve the problem of fast head moving,nonlinear eye tracking and facial fatigue expression detection,a new scheme of driver fatigue detection was proposed based on unscented Kalman filter(UKF) and eye tracking.Owing to the intuition that it is easier to approximate a probability distribution than to approximate an arbitrary nonlinear function or transformation,nonlinear eye tracking can be achieved using unscented transformation(UT) by adopting a set of deterministic sigma points to match the posterior probability density function for eye movement.Driver fatigue can be detected by calculating PERCLOS(percentage of eyelid closure over the pupil over time) under a realistic driving condition after nonlinear eye tracking.The experimental results show that the proposed scheme can not only improve the robustnesses of the head rotating and fast head movement of a driver and the interference of external illuminations,but also provide more accurate estimation than the traditional Kalman filter.
-
Key words:
- eye tracking /
- fatigue detection /
- UKF(unscented Kalman filter)
-
ZHANG Zutao,ZHANG Jiashu.Driver fatigue detection based intelligent vehicle control[C] //Proceedings of the 18th International Conference on Pattern Recognition (ICPR'06).Piscataway:IEEE Press,2006:1 262-1 265.[2] 刘志强,赵艳萍.中国交通安全技术分析[J].中国安全科学学报,2003,13(8):18-21.LIU Zhiqiang,ZHAO Yanping.Technical analysis of traffic safety in china[J].China Safety Science Journal,2003,13(8):18-21.[3] ERIKSSON M,PAPANIKOLOPOULOS P.Eye-tracking for detection of driver fatigue[C] //Proceedings of the international Conference on intelligent Transportation System,Boston.Piscataway:IEEE Press,1997:314-319.[4] HORNG Wenbing,CHEN ChihyYuan,CHANG Yi,et al,Driver fatigue detection based on eye tracking and dynamic template matching[C] //Proceeding of the 2004 IEEE International Conference on Networking,Sensing &Control,Taipei.Piscataway:IEEE Press,2004:8-12.[5] ZHU Zhiwei,JI Qiang.Combining kalman filtering and mean shift for real time eye tracking under active IR illumination[C] //Proceedings of theInternational Conference on Pattern Recognition.Piscataway:IEEE Press,2002:1 051-4 651.[6] 郑培,宋正河,周一鸣.基于PERCLOS的机动车驾驶员驾驶疲劳的识别算法[J].中国农业大学学报,2002,7(2):104-109.ZHENG Pei,SONG Zbenghe,ZHOU Yiming.PERCLOS-based recognition algorithms of motor driver fatigue[J].Journal of China Agricuhural University,2002,7 (2):104-109.[7] 杨秋芬,胡豁生,桂卫华.基于Gabor滤波和特征三角形的驾驶员面部跟踪[J].计算机工程与应用,2006,42(23):16-18.YANG Qiufen,HU Huosheng,GUI Weihua.Driver's face tracking based on feature triangle and gabor filter[J].Computer Engineer And application,2006,42(23):16-18.[8] 陈艳琴,罗大庸.基于Kalman滤波和mean shift算法的人眼实时跟踪[J].模式识别与人工智能,2004,17(2):173-177.CHEN Yanqin,LUO Dayong.Real time eye tracking based on Kalman filter and mean shift algorithm[J].Pattern Recognition and Artifleal Intelligence,2004,17 (2):173-177.[9] PILUTH T.ULSOY A G.Identification of driver state for lane-keeping tasks[J].IEEE Trans.on Systems,Man,and Cybernetics-Part A:Systems and Humans,1999,29(5):486-502.[10] VIOLA P,JONES M J.Rapid object detection using a boosted cascade of simple features[C] //Proceedings of the International Conference on Computer Vision and Pattern Recognition.Piseataway:IEEE Press,2001:1-9.[11] PAPAGEORGIOU P,OREN M,POGGIO T.A general framework for object detection[C] //Proceedings of International Conference on Computer Vision.Piseataway:IEEE Press,1998:555-562.[12] VIOLA P,JONES M.Robust real-time face detection[J].International Journal of Computer Vision,2004,57 (2):137-154.[13] JULIER S J,UHLMANN J K,DURRAN-WHYTE H F.New approach for filtering nonlinear systems[C] //Proceedings of the American Control Conference,Washington,Piscataway:IEEE Press,1995:1 628-1 632.
点击查看大图
计量
- 文章访问数: 1686
- HTML全文浏览量: 88
- PDF下载量: 658
- 被引次数: 0