基于ukf非线性人眼跟踪的驾驶员疲劳检测.pdfVIP

43 6 Vo.l 43 No. 6 2008 12 JOURNAL OF SOUTHW EST JIAOTONG UN IVERSITY Dec. 2008 : 0258-2 24( 2008) 06-069 -06 UKF 1, 2 1 张祖涛 , 张家树 ( 1. , 610031; 2. , 610031) : , UKF. , UT, S igm a, . , PERCLOS , . , , K alm an . : ; ; UKF : TP391 : A Driver FatigueDetection Based on Unscented Kalman Filter and EyeTracking 1, 2 1 ZHANG Zutao , ZHAN G J iashu ( 1. School of M echan ical Eng. , Soutw est Jiaotong Un iversity, Chengdu 610031, China; 2. School of Inform ation Science and Technology, Soutw est Jiaotong Un iversity, Chengdu 610031, Ch ina) Abstract: In order to resolve the problem of fast head moving, nonlinear eye tracking and facial fatigue expression detection, a new schem e of driver fatigue detectionw as proposed based on unscented K alm an filter ( UKF ) and eye tracking. Ow ing to the intu ition that it is easier to approxmi ate a probab ility d istribution than to approxmi ate an arb itrary nonlinear function or transform ation, nonlinear eye track ing can be ach ieved using unscented transformation ( UT) by adopting a set of determ inistic sigm a points tomatch the posterior probability density function for eyemovement. Driver fatigue can be detected by calculating PERCLOS ( percentage of eyelid closure over the pup il over tmi e) under a realistic driving condition after nonlinear eye track ing. The expermi ental results show that the proposed scheme can not only mi prove the robustnesses of the head rotating and fas