Implementation Of Intelligent Face Recognition Surveillance System Based On ARM

With the improvement in the quality of life,family safety has become one of the most important aspects of life.Many places are equipped with a monitor,in order to keep the property safe.So that surveillance products has a huge potential for development.As intelligent video analysis technology moves forward,the intelligent surveillance system has been the trend.The core of intelligent surveillance system is to detect,track,recognize the target,then understand and describe its behavior by using computer vision technology.However,the existing intelligent video surveillance system is mainly based on PC platform.It is high cost,and inconvenient to installation.In this paper,A face recognition intelligent surveillance system based on ARM is Designed and implementing.The system can recognize the face in the monitoring area,and automatically trigger the alarm function if a stranger broken into the family.The main work of this paper are as follows:(1)Because of it is limited computing power,the ARM platform can't quickly recognize faces in high resolution video images.To solve this problem,we propose a face recognition algorithm based on moving object detection.First,we use the ViBe(Visual Background Extractor)algorithm to detect the moving object which appeared in the monitoring area,and then use the AdaBoost and LBP(Local Binary Pattern)algorithms to realize the face detection and recognition in the detected object area.Experiments show that the algorithm not only reduces the computational complexity,but also achieve face recognition in a longer distance.(2)In this paper,when the moving target is detected,we will compute the maximum circumscribed rectangle of the target,and the face detection and recognition are only carried out on the area where the rectangle is located.The experiment found that in many case,the rectangle can only surround the human body,but the face is excluded.Which will largely increase the unchecked ratio.To solve this problem,The paper moves up the rectangle and process only the top center part of images.It can improve the real-time capabilities and the efficiency of the recognition.(3)Design and implementation of the face recognition intelligent surveillance system.The system uses OK210 development board based on Linux operating system as hardware development platform.The system functions mainly include video image acquisition,moving target detection,face detection,face recognition,abnormal alarm,picture preservation and Android client design.It can automatically recognize the face in the surveillance.If a stranger broke into,the system will start abnormal alarm module and remind the user to take precautions.The key images which include the stranger will also be keep.Experiments show that the system is functional and stable.Thus,an integrated solution is made for intelligent surveillance system.