Rsesarch On Intelligent Attendance System Based On Face Recognition Technology

Due to the uniqueness and stability of face images, face recognition has been widely applied in criminal investigation, video surveillance, facial expression analysis, daily attendance and other occasions, and has gradually become a common means of authentication in people’s work and life. Applying face recognition technology into attendance system can make full use of the existed face database to make the authentication more intuitively and easily. But the real face recognition attendance system will face a number of challenges, it needs to provide artificial intelligence compensation to the illumination variation, posture change and the face accessories. In this paper, the face recognition technology for attendance system are studied.This paper has mainly completed the work listed below:1. The acquisition method of face samples is discussed. After implementing the face detection algorithm which is based on haar feature and adaboost cascade classifier, an eye location method based on RNDA feature has proposed. And then normalized face images according to the precise coordinates of eyes.2. The feature extraction algorithms based on LDA is studied. After comparing the common methods which solving the LDA’s small sample size problem, a two-stage LDA face extraction algorithm based on2D-PCA has proposed. The proposed method can solve the small sample size problem and make full use of the four information space of LDA.3. It has studied the classifier which is based on SVM. Firstly, it has discussed the advantages and disadvantages of SVM and NDA, then combined the the discriminatory information represented by normal vectors to the decision surface and the support vectors which are critical for accurate classification, an SVM+NDA classification model is proposed.4. According to face detection, eye location, face feature extraction and classification algorithms, the intelligent attendance system based on face recognition technology has been designed and implemented. The main models of the system are employee information manage, attendance rule setting, attendance information query and electric attendance model.Studies and large numbers of instances show that when the algorithms proposed in this paper apply to the real-time face recognition system, the quality of face samples and the accuracy rate of face recognition have been improved.