基于图像识别的防瞌睡系统设计

Design of Anti-drowsiness System Based on Image Recognition

系统利用Python OpenCV进行图像分析,确定人脸位置并实现人眼定位,利用Dlib库确定人眼的坐标位置。利用脸部局部状态信息进行非接触式测量,计算眼睛纵横比,即EAR值,根据设定的疲劳度阈值比较判断值班人员是否瞌睡。当检测出值班人员瞌睡时,系统会自动报警。实验结果表明,该检测方法使用非接触式测量值班人员生理响应特性,识别精度比较好且满足实用性要求。

The system uses Python OpenCV to analyze the image,determine the position of human face and realize the location of human eyes,and uses Dlib library to determine the coordinate position of human eyes.Non-contact measurement was carried out by using the local state information of the face to calculate the aspect ratio of the eyes,i.e.,EAR value,and judge whether the person on duty is sleepy according to the set fatigue threshold.The system will give an automatic alarm when it detects that a person on duty is drowsy.The experimental results show that the detection method uses non-contact method to measure the physiological response characteristics of sentinels,and has good identification accuracy and meets the practical requirements.