Research Of Adaboost Algrithm In Face Detection And Recognition

Nowdays, with quickly development of computer technique, Face detection and recognition are widely used in the fields of computer vision and safety.,and at the same time more and more peple dedicate to studying face recognition. In this paper,we had made deeply research in algorithm and experiment principle, and On the basis of the study, we finally come up with a new advanced Adaboost algorithms and used the new algorithms in face recognition.Firstly we study and compare three classical recognition algorithms including Principle Component Analysis, Linear Discriminant Analysis and Independent Component Analysis. Also we do many experiments and it proved that the classical algorithms have good performance.Then, on the base of depth study of the application of Adaboost algorithm in face detection. We we used Haar features as weak classifiers and then use Adaboost boost these weak classifiers. a large number of experiments shows that the speed of Adaboost algorithm have great advantage in face detection including fast detection speed, good stability, and good detection performance.Finally, in this paper we study and combine Adaboost.M1 with nearest neighbor algorithm, linear discriminant analysis, Bayesian three weak learning algorithms. after many experiments validation, then we advance a new theory of Adaboost.M1 weight improved algorithm. a large number of experiments show that the the new algorithm accuracy and speed increase. The recognition performance have a great advance compared with before. It is deserve the future research..