Candidate Identity Verification System Based On Face Recognition

In order to ensure the fairness of the test and prevent the occurrence of the surrogate test-taking phenomenon,the identity of the candidate must be verified before the test.Now,in most cases,the candidate identity information is usually authenticated by invigilators.However,when it comes to numerous candidates or intentional alteration on the admission tickets,careless invigilators may not succeed in distinguishing the correct one.Face recognition technology,one of the hot technologies in the world,can be applied to the authentication field.Facing these issues,a method of candidate authentication based on face recognition is proposed to improve the accuracy of candidate authentication.And the main work of this thesis is as follows.The basic principles of image preprocessing and face detection algorithm are studied.The face of the captured image is detected and segmented by using the Vola-Jones algorithm,and the effectiveness of the algorithm is verified by experiments.For the problem that the small sample identification cannot obtain the within-class information,the sparse variation dictionary learning(SVDL)algorithm is used to recognize the face image,Regretfully,the original algorithm also will be influenced by the surrounding environment,lighting,expression,and pose.After studying different facial features,the Gabor features are found to be insensitive to light and can tolerate some rotation.Therefore,a Gabor feature-based sparse variation dictionary learning(GSVDL)algorithm is proposed.Gabor feature is added to the original model to obtain a more accurate sparse representation of the face.And comparison experiments are performed on the face database.The results show that the improved algorithm obtains a higher recognition rate than the traditional face recognition algorithm.According to the research contents above and the examination management requirements from the Academic Affairs Office,the overall framework of the system is completed,as well as the hardware and software design.45 face images of volunteers under different lighting and facial expressions are collected,and a QT graphical user interface is finally established to implement the authentication.