Design And Impiement Of Attendance System Based On Face Recpgnition

With the development of technology, attendance system in enterprises andinstitutions is increasingly and widely used, especially the body biometric attendancesystem which is very popular. In this paper, face recognition based on real-time videostreaming designed for attendance system is adopted.First of all, the system proposed a set of processes used to accurately extract theface region.first, use the algorithm of AdaBoost to detect face image in real time. Andthen through the preprocessing of video frames and color modelings which combine theface images to accurately locate the face. Finally, use Camshift track the located face.Then, a new processing method is presented against the light conditions of faceimages. First, divide the illumination variation into the process of light intensity and theprocess of illumination angle; Second, use the preprocessing of gray normalization toreduce the sensitivity of the light intensity; Third，use five basic point light source toapproximate the light conditions of face recognition application and input the mostclosest light conditions of face images. Based on this, the concept closest illuminationratio image (CIRI) is proposed which is designed to reconstruct standard face imagesand reduce the effect of illumination angle. Establish embedded hidden Markov model(EHMM) for light estimation and reconstructed face image, and then extract itsobservation vector and compare maximum likelihood estimation to the trained datawhich are in the database to determine the identity of the face.Finally, achieved prototype attendance system based on face recognition, Itcompletes automatic input of employee’s attendance information, management andother functions. Through the improvement of this prototype system to meet differentapplication requirements.