摘要
根据监测系统车载、实时、非接触、小型化等要求,设计并实现了一种基于ARM+Linux的嵌入式实时疲劳监测系统。将图像数据预处理、算法参数优化、汇编优化和编译器优化等优化方法相结合,实现了快速的人脸检测,并通过移植OpenCV实现了基于PERCLOS的人眼疲劳检测算法。与基于PC和DSP的疲劳监测系统相比,该实时疲劳监测系统具有强大的网络功能以及更优越的扩展性。在室内环境下的实验结果表明,系统检测准确率达到95%。
According to the factor of vehicle, real-time, non-contact, small size, an ARM＋Linux-based real-time fatigue monitoring system is designed and implemented. The fast face detection is implemented by means of integrating methods of image data preprocessing, the algorithm of the parameter optimization, assembly optimization, compiler optimization. The eye fatigue detection algorithm based on PERCLOS is achieved by migrating OpenCV. Compared with DSP-based and PC-based fatigue monitoring system in present, this real-time fatigue monitoring system has powerful network feature and better scalability. Experiment results demonstrate that the correct rate can reach 95% in indoor scene.
出处
《电视技术》
北大核心
2011年第13期106-109,共4页
Video Engineering
基金
山西省自然科学基金资助项目(2009011023)
