智能监控系统面临的主要问题是视频监控图像数据量大,对实时性传输、存储和处理的要求高。如果将智能监控中的人脸识别的工作前移,即利用前端嵌入式摄像头设备实现人脸识别,不仅可以实现人脸快速检测以及危险目标出现时的实时预警,还可以大大降低智能监控人脸识别系统对通信、存储和计算等性能的要求。本文基于树莓派实现了一个人脸识别原型系统,验证了该方法的有效性。

The main problems of an intelligent video surveillance system are the real-time transmission,storage and processing of large amount of video data.If the face recognition work is moved forward,that is to use front-end embedded equipment to do face recognition,not only rapid face detection can be implemented,but also performance requirements for communication,storage and computing can reduced greatly.In the paper,a face recognition prototype system is designed based on Raspberry Pi to verify the effectiveness of the method.