【摘要】 司机疲劳驾驶实时检测系统在实际应用中有很重要的意义。设计了一个利用图像分析的方法,通过测量PERCLOS指标值来进行疲劳判断的该类系统。系统首先对图像进行预处理,然后采用基于YCbCr颜色空间肤色模型进行人脸粗定位,根据人脸特征,逐次进行人眼区域缩小;最后通过对边缘信息进行先验知识结合积分投影的方法进行人眼定位和闭合度测量。考虑到视频图像序列帧与帧之间的相关性,采用线性运动预测的方法对人眼进行跟踪,减少了系统的运算量。实验结果表明系统能实时、准确地反映司机的疲劳状态。更多还原

【Abstract】 Real-time driver fatigue detection system has a great value in practical application.A new system based on PERCLOS detection by the ways of image analysis is designed.Firstly,image is processed before face area location using skin color model in YCbCr color space which is in order to decrease eyes search area. Then,eyes are located by integration together with prior information and the degree of eyes opening is detected. Computation complexity of the system is decreased,because eyes are tracked by linear motion forecasting for many correlations among frames in video images.Experiments result shows that this system could reflect driver’s fatigue degree in real-time and precisely.更多还原