| With the development of science and technology,the awareness of safety and defense is also increasing as people pursue a high level of life.As the basis of security,the access control system has important research significance.The traditional access control system is based on password or certificate authentication,which is gradually eliminated due to the shortcomings of weak security and complicated operation.With the rapid development of biometric technology,face recognition has been widely used in image processing,video surveillance,and other industries due to its advantages of non-contact,non-interference,simplicity,and convenience.Therefore,the access control system based on face recognition has high safety factors and reliability and is favored by most consumers due to its automation and intelligence.Traditional face recognition access control systems are based on high-performance large computers with massive storage capacity.It can show a good recognition effect on large occasions with large personnel changes.However,due to its large size,high cost,and insufficient portability,it has no good market benefits in private places such as homes and laboratories.With the development of the embedded system,the performance of an embedded processor has been greatly improved.Therefore,this dissertation proposes the design of a small and medium-sized face recognition access control system based on an embedded system and elaborates on three aspects: algorithm analysis,hardware design,and software design.In terms of algorithm analysis,image preprocessing operations are needed to improve the image quality,including image format conversion,image grayscale,histogram equalization,median filtering,and size normalization.Adaboost algorithm based on Haar feature is selected for face detection.After determining the key information of the face according to the Haar feature,the eigenvalue is calculated by the integral graph method and sent to the cascade classifier cascaded by several strong classifiers trained by the Adaboost algorithm for face detection.Through the analysis of the three face recognition algorithms of PCA,LDA and LBPH,the algorithm based on LBPH is finally selected for face recognition.In terms of hardware design,the system adopts the Exynos4412 processor as the main control module.According to the functional requirements of the system,the peripheral circuits of the system are designed one by one.To meet the power supplydemand of each part of the system,5V,4V,3.3V,and 1.8V power supply circuits are designed;To solve the problem of insufficient memory of the processor,the SDRAM interface circuit and e MMC interface circuit are designed;To carry out humancomputer interaction,the interface circuits of LCD display screen,capacitive touch screen,and USB keyboard are designed;To meet the communication requirements,UART,JTAG,and Ethernet interface circuits are designed;To collect images and control access control,the camera interface circuit and electromagnetic lock control interface circuit are designed.In terms of software design,choose to build an embedded Linux operating system development environment on this target platform,and transplant the Qt library of the graphical user interface and the computer vision library Open CV.According to the system’s functional requirements,the system main program,image acquisition program,image preprocessing program,face detection program,face recognition program,data storage program,and electromagnetic lock driver are designed.Finally,the function of face recognition is successfully realized on the embedded platform.The test of the system’s face detection and face recognition functions shows that the system has high recognition speed and accuracy,and is easy and fast to use when the data sample is small.Compared with the PC,this system has the advantages of lower cost and better expansibility.The operation effect in small and medium-sized places in real-time,stable,and reliable,and can achieve the expected goals.Figure [48] table [6] reference [56]... |
PDF Full Text Request