Research And Implementation Access Of Control System Based On Face Recognition

With the development of the times,intelligent door locks that incorporate human biometric features are increasingly favored by people.Compared with other biometrics,facial features are intuitive,simple,accurate,and non-contact,so it becomes the mainstream recognition feature of access control systems.The versatility and cost performance of the existing face recognition technology access control system can not meet the requirements of the civilian market,and it is urgent to improve the recognition rate under unfavorable lighting conditions and different steering situations.Based on machine learning algorithm and combining Android mobile phone authentication and embedded development technology,this paper designs a low-cost and efficient access control system for face detection and identification for small office applications.The main work involved is as follows:For the access control camera,the face is susceptible to light.The access control system needs to quickly detect the face and simultaneously detect the requirements of multiple faces.The Adaboost cascade classifier based on multi-scale block LBP features is trained.The multi-scale block LBP feature is based on the gray average of each sub-region and is more robust than the LBP feature.Experiments on a self-built database show that the multi-fork stump weak classifier based on multi-scale block LBP features has a high detection rate under the above conditions,and can accurately calibrate the position of the face.The weak classifier constructed by the set threshold method used by the conventional Haar feature has a higher detection rate and is faster.For the redundancy of face image data captured by the access control camera,the principal component analysis(PCA)algorithm is used to reduce the dimension of the face image,and the projection vector of the feature face subspace is used to express the face sample in the low dimensional space.Because the traditional BP network converges slowly and is easy to fall into the local minimum point,the weight adjustment method is improved.The improved BP network classifies the samples,which greatly shortens the training time and significantly improves the performance of the BP network.In this algorithm,in order to reduce the false detection rate and improve the face recognition rate,the face invariant feature is added for secondary verification.Experiments on the reorganized face database,the recognition rate of the improved face recognition algorithm in this paper is 97.5%.In the hardware design of the access control system,the development board is used as the server of the access control system;Huawei glory V10 is used as the mobile phone client;the color industrial camera is used as the camera,which has the advantages of low cost and low power consumption;and has a 12-megapixel high-resolution image sensor.This paper develops an access control system based on face recognition on the Android platform,and implements functions such as user registration,user login,execution of door opening and closing instructions,and query execution log on the mobile phone client.The operation is convenient and simple,and is suitable for small office places.The server side is transplanted into the face recognition algorithm,which can quickly and correctly detect and recognize the face image captured by the camera.Data transmission,selection of custom transport protocols,and encryption of data using the Advanced Encryption Standard(AES)symmetric encryption algorithm to improve system security.In addition,the system's function and performance have been thoroughly tested.The test results show that the system can meet the requirements of the access control system.