摘要
文章提出了一种在实时视频图像中快速检测人脸并检测司机工作状态的的方案。利用VC设计了软件将其实现。对摄像机采集的日间图像,先进行光线补偿和中值滤波,然后采用肤色检测定位人脸,最后运用双眼模板判断司机的工作状态;对于夜间图像,采用差分方法进行处理。实验证明该方法快速,准确率高,有效适用于实际工作环境。
A face detection method that can be applied to real Ctime video-based working state of motorman monitor system is proposed. And it is realized by Visual C＋＋. Firstly, the images of day that captured by camera were proceed by light compensation and filter. Secondly, skin color was used to detect the face from the region. Thirdly, two-eye template was used to judge the state of the motorman. While, every two images of night were subtracted to judge the state of the motorman. Experiments with variety of images show that the method proposed here is efficient and useful.
关键词
司机工作状态
人脸检测
肤色模型
模板匹配
Working state of the motorman
Face detection
Skin color
Template matching
