Research And Development Of Multi-person Identification System Based On Adaboost Face Detection Algorithm

With the rapid advancement of the digital information age of human beings,the requirements of information security in modern social life are getting higher and higher,and the identification technology of biometrics occupies an increasingly important position,in which face recognition is in identity verification and face.Identification and face payment have important implications,and the role of multi-face recognition in video scenes is increasingly important.Face recognition technology is widely used in all aspects of society,such as public monitoring,face attendance,and smart education.This paper takes the university classroom video as the background,carries out the research of the key algorithm of the multi-person identification system and the realization of the system.Through the multi-person identification system,the face recognition of the university classroom video is realized,and then the classroom real-time head-up rate,the average head-up rate of the classroom and Statistics on student classroom head-up rates.The head-up rate can reflect the real situation of the university classroom,and the quality of classroom teaching can be evaluated.This is of great help to the reform of education and teaching quality in colleges and universities.In this paper,multi-face detection algorithm,multi-face tracking algorithm and multi-face recognition algorithm are studied,and the development of multi-person identification system is carried out.Firstly,the traditional Adaboost algorithm is improved in the research of multi-face detection algorithm.The improved algorithm and skin color model detection algorithm are combined to form a new multi-face detection algorithm.Kalman filtering is applied in the research of multi-face tracking algorithm.The Camshift tracking algorithm is used to estimate the face position to realize multi-face tracking in complex scenes.The multi-face recognition algorithm researches the face features extracted by LBP to reduce the dimension of PCA,speed up the recognition and realize multi-person Quick recognition of the face;using Matlab computer-aided software,Opencv computer vision library and MIT,ORL,Yale standard face database to verify the algorithm before and after the improvement.Finally,the implementation of multi-person identification system was carried out using Visual Studio 2012,OpenCV,Qt and MySql database.The functions of the system were tested one by one,and the performance of the system was tested in real time,accuracy and CPU usage.Through experimental verification and system testing,the multi-person identification system can achieve a recognition rate of 91.6% in the college classroom video scene.The system application can perform real-time head-up rate,classroom average head-up rate and student classroom head-up rate statistics under the classroom video.Sexual effects meet the requirements.