Design And Implementation Of Identity Authentication System Based On Face Recognition

Identity authentication is an important part of system security.If the staff who log in the system are not consistent with the staff who operate the system,it will bring immeasurable losses to the enterprise,therefore,to ensure the consistency of identity authentication is the premise to ensure the safe operation of the system.It is common to operate the system instead of others,and it also appears that a staff replace another staff to login system using card reader easily,compared to the first two ways,biometrics is relatively safe and has few phenomen instead of others operating system.Currently,the main biometrics technologies used in the login system are fingerprint recognition and face recognition,because fingerprint recognition uses contact recognition and the recognition rate is susceptible to the influence of finger skin,face recognition uses non-contact recognition,which is more acceptable than others.The main contents of the thesis are as follows:1.In accordance with the consistency between the staff who login in system and the staff who operate the system can not be guaranteed.Therefore,the photo deception detection is needed before face recognition,and the photo deception algorithm based on the LBP features is proposed for liveness detection.Finally,according to the classification result,the susquent operations are proceeded.2.In accordance with the low accuracy of the traditional eye location classifier under the condition of wearing glasses in the face image preprocessing,it is necessary to detect the glasses before the eye location,and then choose the different eye locator according to the result of the glasses detection.Using the morphological glasses detection method,the face is segmented according to the input gray-scale face image and the roi of the glasses border is obtained.A series of morphological operations are carried out on the region of interest.Finally,the edge detection is carried out.It improves the accuracy of eye location and the rate of face recognition.3.The rotation angle is calculated according to the eye location and the image is rotated by affine transformation.The rotated image is apllied to face recognition to improve the face recognition rate.CNN algorithm is used to recognize the face and identify the classification information.The system has been applied in enterprises and has good application effect.