Research On Dormitory Access Control System Based On Face Recognition

Identity authentication technology is the core of the access control system.Due to the rapid development of biotechnology,computer technology and information technology,identity authentication technology has undergone revolutionary changes.Traditional authentication is mostly a magnetic card,a digital password,etc.,and there are many drawbacks.Nowadays,advanced identification methods mostly use biometric identification technology.Among them,face recognition has attracted the attention of scholars and scientific research institutions because of its user-friendliness and low equipment requirements.In recent years,the development of deep neural network technology has injected new blood into the field of face recognition.Based on the theory of convolutional neural network,this paper implements a dormitory face recognition access control system,which has certain security,and has achieved good results,and has high practical value.The main contents of this study are as follows:1.Analyze the development of face detection technology,adopt a multi-level convolutional neural network structure for face detection,set three different loss functions according to the network learning task,and finally obtain the model in the mainstream face image set.A higher detection rate can be achieved.For the use environment of this study,the histogram equalization method is used to optimize the nighttime imaging,which can effectively improve the influence of insufficient light.The median filtering method is used to control the image noise,which can improve the accuracy of detection and recognition to a certain extent.2.Improved feature extraction network is used to extract features from faces.The improved network deepens the network depth while having a lightweight parameter quantity,which can improve the running speed and classification performance.The resulting model achieved good results on the mainstream face image set.3.A blink detection method based on eye feature points is proposed.Compared with the traditional blink detection method,the proposed method has fast running speed,simple calculation and high accuracy.4.The WEB-based dormitory face recognition access control system is implemented,which includes functions such as personnel registration and authentication.The system has low requirements on equipment and strong practicability.The system adopts the model obtained from the training of this research.It has fewer parameters and higher precision.It is suitable for the use of access control system and can improve the efficiency of access control system.