Research And Implementation Of Intelligent Attendance System Based On AdaBoost Algorithm

With the development of modern society and technology, the requirements of the reliability of identification for companies and enterprises are highly increasing. As a new kind of identification technology which combines the computer technology with some high-tech methods including biosensor, optics and acoustics, biological recognition technology is drawing people’s attention gradually. The face recognition technology is one of biological recognition technologies which have a widely application. This thesis first investigated the face detection technology and the face recognition attendance system. Then, it constructed an intelligent attendance system based on OpenCV and AdaBoost algorithm.In the process of constructing the system, we used AdaBoost algorithm and face database to train the system and xml parameter file for face detection. Then we apply the OpenCV library function for video capture, image processing and face recognition. Finally, the attendance can be finished. The main works of this thesis are as follows:(1) The key techniques of intelligent attendance system were studied and the system architecture was given. The methods of face detection and face recognition were discussed.(2) The research and realization of AdaBoost algorithm were presented in this thesis. We trained the similar Haar feature to the weak classifier by the integral image method, and then several weak classifiers formed a strong classifier according to a certain weight. At last, we cascaded several strong classifiers into a cascade classifier. We trained on the ORL face database with a cascade classifier, whose output could be used for face detection.(3) According to the demand of the attendance system, we divided the system into several modules, and realized the functions of the modules. Our system can capture the video by video camera and display the collected video stream in the window. At the same time, it could detect the video stream by the cascade classifier which is obtained through AdaBoost algorithm, and mark the detected faces using rectangle. Furthermore, it could preserve the collected faces when the faces reach the suitable size and identify the check-in staff via comparing the collected face with staff face database.(4) The whole system based on MFC/C++ was implemented and tested. We created a login interface and attendance interface to the face recognition attendance system and realized the bottons and other controls’ response. After completing the system test, we found some bugs and system flaws of the software and got rid of them.