The Design And Implementation Of The Intelligent Entrance Guard And Attendance System Based On Face Recognition

The entrance guard and attendance system is an important part of the security system. The traditional mechanical systems always adopt artificial registration, mechanical lock or brushing card, etc. However, these operations which involve many artificial factors, such as loss, theft, or manually close operation method and so on. Therefore, these operations are complex, and unreliable. With the rapid development of computer technology and Internet, intelligent system based on human biometrics is very active in recent years. It has become a research hotspot in the field of the computer vision and the pattern recognition.Now, the intelligent system based on human biometrics has successfully applied some biological characteristics into the traditional entrance guard and attendance system. And it overcomes the shortcomings of the traditional manual way that is complex and unreliable. However, the intelligent system generally uses the biological characteristics of the fingerprint and the iris. Using these characteristics always asks for special acquisition device, and close contact. Besides, there are some problems like easy to be deliberately falsified, or difficult technology.As face is a special biological characteristic with uniqueness, the face recognition technology can be used to the entrance guard and attendance management. This fully embodies the advantages of convenient operation, high performance, precision, reliability and not easy to forge. However, the common problems are that the processing speed is too low, and that it only supports the single target processing. Besides, it requires too much on people's behavior. As multiple targets appear at the same time or stay for a period of time, the traditional methods may cause pedestrian counting chaos and identification errors. Thus, it can't satisfy the requirement in practical application.To cure the above problems, this paper presents a series of strategies. The system uses the coincidence degree for clustering, and then identified multiple targets by using the PCA and LBP recognition, according to the similarity constraint. If the similarity constraint does not meet the threshold requirement, the system can use the deep learning to identify the targets again. Experimental datas show that these strategies can make the systemperformance better, but also can obviously increase the recognition rate in multiple targets recognition.In this paper, an intelligent entrance guard and attendance management system is designed and implemented, based on the existing research achievements in the field of face recognition. The system adopts the Qt graphical user interface(GUI) tools which can be used across platforms. While establishing the platform system, the lightweight Sqlite database was adopted. At the same time the system used the socket communication to realize the communication between the client and server. The system consists of three subsystems: real-time analysis system, display system, attendance statistics system. The real-time analysis system is responsible for the real-time face recognition and pedestrian traffic statistics. The display system displays the information about the arriving person and prompts the voice. The attendance statistics system can carry out the records within a certain time period in accordance with the requirements of related queries, but also will result in the form of Excel export. Experimental results show that the system using these strategies can solve the bottleneck problems of multi-target recognition and the pedestrian traffic statistics, also the video processing speed can meet the real-time requirements.