基于PERCLOS的眼睛张开程度检测算法研究

Algorithm research of the eye open degree detection base on PERCLOS

针对疲劳驾驶检测中驾驶员眼睛状态如何确定问题,文中提出一种眼睛张开程度检测算法。该算法依据虹膜(含瞳孔)暴露区域面积反映眼睛张开程度。文中首先检测虹膜左右边界,根据左右边界高度以及边界之间距离判断眼睛开闭情况。接着对开眼图像检测虹膜与上下眼睑交界,计算虹膜暴露面积。将其与眼睛长度的比值作为眼睛张开程度值。利用统计方法确定PERCLOS评测所需参数。试验结果表明该方法具有定位简单准确,对光照变化不敏感,检测眼睛张开程度精确等优点。

Aiming at the eye＇s state in driver fatigue detection,an algorithm to detect the eye open extent is proposed.This algorithm depends on the size of iris（including pupil） exposition area to reflect the eye open information.In this paper,first,detect both left and right boundaries of the iris,and judge the eye open or closed based on the height of both left and right boundaries and the distance of two boundaries.Then,detect the junction of the iris and both upper and lower eyelids,and count the iris exposition area.The ratio of the area and the eye＇s length is the value of eye open degree.PERCLOS parameters are determined using statistical method.Experimental result shows that this method is a simple and accuracy orientation way,is not sensitive to light,and can detect accurately eye open degree.