基于视觉的疲劳驾驶检测算法

An algorithm of fatigue driving detection based on computer vision

在疲劳驾驶检测中,眉毛、眼睛等相似物体常常会引起误判。为了减小误检率,文章提出了一种基于扩展Haar-like特征的AdaBoost算法,并将其应用于对人眼的准确检测中。通过近红外摄像头获取驾驶员的脸部图像,减少光线对检测方法的影响;利用AdaBoost算法直接检测睁眼区域;计算连续闭眼的帧数占总帧数的比值,判定驾驶员的疲劳状态。在白天、夜晚以及光线突变的条件下测试的结果表明,该方法准确率较高,并且在光照和人脸角度变化的条件下,能准确地定位人眼区域,对其进行疲劳检测。

In fatigue driving detection, the similitude of eyes and other objects can always cause the misjudgment. In order to reduce the false detection rate, an AdaBoost algorithm based on the extended Haar-like features is proposed. Firstly, the facial images of drivers are captured by the IR CCD camera to reduce the effect of illumination. Then the open eye region is located by AdaBoost algorithm. Finally, the ratio of consecutive shuteye frames to total frames is calculated to judge the driver＇s fatigue status. In real driving conditions, the method is verified in the daytime, at night and in different lighting conditions. The experimental results show that this approach can accurately locate the eye region and detect the fatigue status under varying light and facial angle conditions.