Design Of Face Recognition System Based On Linux

With the continuous progress of the society and the urgent demand for fast and effective automatic identification in all aspects,face recognition technology has developed rapidly in the last few decades.With the continuous improvement of the accuracy of face recognition algorithm,face recognition has been widely applied in many fields such as security,finance,e-government and so on.Automobile driver training management system is one of the most popular applications.This paper designs an embedded face recognition system aiming at the problem that the automobile soldiers are prone to cheating in the course of driving training.It is effective to eliminate the phenomenon of teaching replacing and lerning replacing.In this paper,the research status of embedded system and face recognition system at home and abroad is investigated.The face recognition system based on ARM9 and Linux operation system is researched and designed.This system selects the jz2440 development board as the hardware platform,its microprocessor is ARM9,and the software platform selects the open source Linux operating system.Firstly,the principles of face detection and face recognition algorithms are discussed in depth,based on the traditional AdaBoost algorithm in the training of the classifier to the training phenomenon,puts forward a face detection threshold in the set weight update algorithm,and the algorithm was tested.The experimental results show that the improved algorithm improves the detection rate and reduce the the detection error rate.In the face recognition algorithm,the algorithms of PCA(Principal components analysis),LBP(Local Binary Patterns)and CNN(Convolutional Neural Networks)are studied in depth,and the performance of the three algorithms is compared through experiments,the experimental results show that the CNN algorithm is robust to face pose.Secondly,this paper introduces how to set up Linux system development platform,installing cross compiler tool chain on the PC,the development environment of Bootloader transplantation,Linux kernel transplantation,root file system transplant and Linux device driver transplantation,provide a stable and reliable environment for the embedded software design of face recognition system.In terms of system design,face recognition system is divided into four modules,including human-machine interaction module,camera acquisition module,face detection module and face recognition module,and each module is designed and implemented.Finally,the facerecognition system is transplanted to the ARM(Advanced RISC Machine)development board successfully.The test results of the system show that the recognition rate of the traditional algorithm is higher than that of the small sample library,reaching 95%,and the recognition rate is faster.The recognition rate of the depth learning algorithm on large database is obviously higher than that of the traditional algorithm,but the recognition speed is relatively slow.This paper has a good reference value for the application of embedded face recognition system.