Design And Implementation Of Face Recognition System Based On Convolution Neural Network

With the continuous development of society,the identity information of people is becoming more and more important in production and life.Face Recognition Technology is not only a hot topic in computer vision research,but also has been widely used in security,finance,e-government and other fields.In this paper,we mainly study the relative application of Convolution Neural Network model in natural scenes,which was used in the field of face recognition with depth learning method.Compared with the traditional method of face recognition,the model of deep Convolution Neural Network does not need people to design the algorithm of feature extraction,which is complex and time-consuming.We only need to select or design an effective neural network model,and to do some simple and efficient training under a large number of training samples,this method can extract the characteristics of the image and get relatively good classification accuracy.The performance and effectiveness of this approach depend primarily on the design of the network structure.Therefore,in this paper,we focus on how to build a reasonable network model and adopt the relevant technology to make it fast and stable in the training set.Ultimately,we also need to get a good recognition effect.In this paper,the methods of face detection and face recognition are analyzed,optimized and realized.In the process of face detection,the Haar feature is combined with Adaboost algorithm,and the integral graph method is used to speed up the calculation of Haar characteristic,which can quickly and efficiently realize face detection.This module not only realizes the function of static and dynamic face detection,but also embeds face detection into face recognition system to improve the efficiency of face recognition.In the face recognition process,the improved VGG network model is obtained by reasonable reduction of the original VGG convolution neural network training parameters,and the convergence time of the model is reduced by using the parameter initialization method which is better than random initialization.Finally,the new model not only solves the original VGG model hardware requirements,training difficulties and other aspects of the problems,and successfully applied to recognize faces in the natural environment.According to the result on LFW,the accuracy rate of this model can reach 92%.In this paper,a real-time face recognition system is implemented by applying the models mentioned before.The functions and processes of each module of the system are introduced in detail,and the self-built face database is used to achieve the accuracy of 94%.The system verifies the effectiveness of the proposed method and achieves the application requirements of face recognition.