基于MBLBP与Haar特征提取的驾驶疲劳检测

DETECTION OF DRIVER FATIGUE BASED ON EXTRACTION OF MBLBP AND HAAR FEATURES

为了提高驾驶员疲劳驾驶的检测效率,提取人脸和人眼的MBLBP与Haar特征,使用gentleboost训练方法得到多级强分类器,用级联的方法对人脸和人眼进行检测和识别.当系统连续三帧检测到人脸但检测不到人眼的情况下判定驾驶员出现疲劳.加入人脸肤色验证信息,构建6个视角分类器,使用动态表观模型(AAM)来拟合人眼.试验结果表明大大提高了驾驶员疲劳的检测效果.

To improve efficiency in detection of driver fatigue, MBLBP and Haar features in human faces and eyes were extract. Strong multi-level classifiers were obtained by the method of gentleboost training, and the method of cascade was applied to detect and identify faces and eyes. With the system detecting face but not eye for three continuous frames, it is determined that the driver is fatigued. With color authentication information added to human face, and six angle classifiers built, the active appearance model （AAM） was used to fit the human eyes. Results showed that the efficiency of driver fatigue detection was greatly enhanced.