Research And Application Of Face Recognition Algorithm Based On CNN

As an important research direction in the field of computer vision,face recognition has a good application prospect in all fields of life,and is the simplest,most convenient and safest biometric technology.At present,the traditional method of face recognition has achieved high accuracy under the condition that the environment is relatively simple and fixed,but in practical applications,the environment is more complicated,and it is easily affected by factors such as illumination,posture and occlusion.The face recognition task encounters certain challenges.Moreover,the existing methods of face recognition based on convolutional neural network are also less robust and computationally complex.Therefore,this paper studies a method of face detection and recognition based on convolutional neural network.And apply it to the face sign-in system.The main work of this paper is as follows:(1)Explain the research background and significance of the subject,and read a large number of Chinese and foreign literatures,understand and summarize the research status of face recognition at home and abroad.At the same time,briefly introduced the basic framework of face recognition and the theoretical basis of convolutional neural network.(2)Aiming at the problems of poor robustness,complex structure and large computational parameters of existing methods of face detection based on convolutional neural network,a face detection algorithm is studied.The algorithm mainly includes two parts,candidate frame generation network and the face classification network,it can not only reduce the parameters,but also improve the robustness of the algorithm.(3)Using the classical VGG network model and simplifying and improving it,this paper studies a method of face recognition based on convolutional neural network.In order to reduce the intra-class gap when the gap between the faces is increased,the softmax loss function and the center loss function are combined for joint training,which improves the accuracy of face recognition.(4)Based on the algorithm studied in this paper,a face sign-in system based on IPv6 is designed and tested and analyzed from three aspects:illumination factor,face gesture and face occlusion on the self-built small face database,verifying the performance of the system.