论文部分内容阅读
疲劳驾驶己成为导致交通事故的重要原因之一,对自动驾驶中的疲劳预测研究已变成现阶段的重要课题。本文就驾驶员的嘴部图像检测来判断驾驶员的疲劳状态。首先,采用Adaboost算法通过人脸分类器检测出人脸部区域,并粗检嘴部区域;其次,分割出嘴部二值图像;最后,利用神经网络对二值嘴部图像进行训练以预测是否疲劳。实验结果表明,对嘴部分割的方法可以有效避免光照对图像分割的不利影响,提高预精度。
Fatigue driving has become one of the important causes of traffic accidents, and research on fatigue prediction in automatic driving has become an important issue at this stage. In this paper, driver’s mouth image detection to determine the driver’s fatigue status. Firstly, Adaboost algorithm is used to detect the face area of face by face classifier and to check the mouth region; Secondly, the binary image of the mouth is segmented; Finally, the binary mouth image is trained by neural network to predict whether fatigue. Experimental results show that the method of mouth segmentation can effectively avoid the adverse effects of light on image segmentation and improve the accuracy.