改进的积分投影眼睛定位算法

Improved Eye Location Algorithm of Integral Projection

基于计算机视觉处理的方式大多都是通过分析眼睛的状态来判断驾驶员的疲劳状态,其中,眼睛的精确定位是关键环节。提出一种改进的积分投影眼睛定位算法,通过在自然环境下先检测出人脸,然后进行眼睛的精准定位。首先利用肤色分割对人脸进行初定位,在初定位的基础上结合形态学框选出人眼区域,然后利用水平和垂直积分投影曲线精确定位眼睛。实验结果表明该算法简便易行,速度快,定位准确度较高,能满足疲劳检测的应用要求。

Most of vision-based methods are carried on through the analysis of eye state to confirm the fatigue state,in which the accurate localization for the eyes is improtant.A kind of improved integral projection eye location algorithm is introduced,which detects the face under natural environment in advance,and then locates the eyes accurately.Firstly,using the skin color division to initially locate the person face and combining the morphology frame to select the human eye region.Then,using the horizontal and vertical gradation integral projection curve to find the eye location.Experimental results prove that the method is simple and easy to be realized,and has higher accuracy and speed performance.In addition,it also meets the application requirement of fatigue detection.