该项目旨在开发并发布一个性能合理的免费软件“运动检测、人脸检测与识别”系统。

The project aims to develop and release a freeware “motion detection, face detection and face recognition” system with a reasonable performance rate. 

该系统应在实时视频输入源上工作，并在无法识别人脸时采取适当的应对措施。

This system should work on a real time video input feed and should also take appropriate actions when a face is not recognised.

这些模块在测试后实现的性能速率与相关研究论文中描述的性能速率大致相同。

The implementation of these modules upon testing yields performance rate approximately equivalent to those established in their respective research papers.

尽管当前的系统需要大约11秒来处理每一帧数据，但可以通过C语言取代matlab®代码来改进这一点，正如许多其它商业软件所做的那样。

Although the current implementation takes around 11 sec to process each frame, this can be improved by implementing the MATLAB® code in C, as many other commercial vendors have done.

如果出现未能识别的面部，则采取两种措施。

In the event of an unrecognised face two actions are taken.

首先是通过电子邮件向移动通信设备发送该面部信息。

First is to email that face to a mobile communication device.

此电子邮件通过互联网从PC发送，并由移动设备通过2G或3G或Wi-Fi连接接收。

This email is sent from the PC via internet and is received by the mobile device via 2G or 3G or Wi-Fi connection.

接收通知所用的时间几乎是瞬时的，但实际上取决于服务器负载和移动设备的收件箱刷新/发送速率。

The time taken for the notification to be received is almost instantaneous but it does depend on the server load and the inbox refresh/push rate of the mobile device. 

第二步措施是将该帧数据添加到视频文件中，用户可以稍后查看该文件。

The second is that the frame is added to a video file which can be later viewed by the user.

完整论文及相应源码下载地址：

http://page5.dfpan.com/fs/clcj3221e29156fd6f8/

更多精彩文章请关注微信号：