Study On Face Recognition Based On Convolutional Neural Network

Face recognition is an important biometric technology.It has the advantages of high reliability,convenience,and so on.It has become a research hotspot in academia and industry,and has been widely used in the fields of identification,public security,and intelligent terminals.In recent years,with the vigorous development of deep learning technology,the face recognition method based on convolutional neural network has achieved remarkable results,and the recognition rate has been greatly improved compared with the traditional methods,which has effectively promoted the commercialization of face recognition systems.However,when actually applied,affected by external factors such as light intensity and light angle,the facial image captured by the image capture device may be degraded,which will seriously affect the effect of face recognition.In addition,most open-source face data are currently dominated by Western faces,which makes the trained model have poor recognition of Eastern faces and is difficult to put into practical use.Based on image enhancement technology and deep neural network technology,this paper studies the technologies related to face recognition.The main work of this paper is as follows:(1)Build an end-to-end face recognition system,including four parts: image preprocessing,face detection,face alignment,face feature extraction and classification.The first three parts can be seen as preparations for face recognition,and feature extraction and classification are the core of face recognition.(2)Study the enhancement process of dark channel theory on fog and low-light images,and apply it to the preprocessing of face images.This method achieves image enhancement under poor illumination,uneven light,and other harsh lighting conditions.The image enhancement method improves the ability of the system to detect human faces,and lays a good foundation for the follow-up work of face recognition.(3)In view of the lack of Oriental face data,using the idea of transfer learning to make full use of the face representation ability of pre-training models,using pre-training models and classifiers to achieve face recognition for the face of the East.Use some data to fine-tune the model,to enhance the recognition of the eastern face.