Research On Face Recognition System Based On Edge Computing

The Internet of things develop rapidly and there are more and more intelligent devices around us,and with them comes a lot of data,which also puts great pressure on the way that cloud computing processes data.The rise of edge computing means processing near the physical data of intelligent devices on the Internet of things,rather than uploading it into the cloud.Edge computing relieves the pressure of data processing by cloud computing,reduces the reaction time of intelligent devices,and ensures the safety of data transmission.In this paper,a face recognition system based on edge computing,which is used for video surveillance of edge terminal equipment,is proposed to process the image at the end of the image produced by the camera,so as to reduce the demand of cloud computing and Network Bandwidth the processing results are returned for further decision-making,reducing the equipment reaction time.This paper introduces face recognition,image fusion and face recognition system construction through face recognition to realize identity authentication.The self-made face data are fused by three methods: spatial domain image fusion,wavelet transform image fusion and curvelet transform image fusion.Face detection,image pre-processing,LBP feature extraction and KNN classification are performed for the unidentified people,and then compared with the fusion image data to realize face recognition.The experimental results show that the fused images obtained by curvelet transform not only get the best image quality,but also effectively improve the accuracy of face recognition in Raspberry Pi.Finally,a face recognition system is built to judge the result of the image processed by Raspberry Pi,and realize the identity authentication.By building and realizing the system,the face recognition authentication can be well realized,and it is cheap and easy to implement.A face recognition system based on edge computation is laid out in primary and middle schools,which can effectively monitor dangerous persons and reduce the probability of occurrence of unexpected events.