Research On Face Detection And Recognition Algorithm

In recent years,biometric technology as an effective means of identification.It has received great concern.Compared with other biometric technologies,face recognition technology has significant advantages.Therefore,it has been widely applied to many fields and played a key role.Therefore,Face detection and recognition technology has gradually become a major field of computer vision area.This thesis concentrates on face detection and recognition,aiming at improving the detection and recognition result.This paper firstly gives a brief description of the research background and significance of face recognition,and expounds the research status of face recognition in domestic and abroad,and then summarizes the research contents and chapter structure.This paper also introduces the key technology of face detection and recognition in order to lay a good foundation for the further research.Also,some improvements are made to the existed algorithms in order to obtain better retrieval results.The main research contents in this thesis are as follows:(1)Aiming at the shortcomings of traditional Adaboost algorithm in the selection of weak classifiers and weights,put forward a new face detection algorithm based on an improved Adaboost.This method first use some support vector machines as weak classifiers.At same time,by combining the differences of different base classif,which can reduce the number of weak classifiers,and improve the efficiency of the algorithm.And then the genetic algorithm is used to optimize the weights of weak classifiers.Finally,using Adaboost algorithm to get a cascade classifier that has been compensative by a number of weak classifiers.The experimental results show that using this improved method can improve the detection efficiency,and shorten comuutation time.(2)Face feature extraction based on improved CS-LBP.To improve the face recognition rate,this paper proposes to use Laplacian filters to stress details or enhance fuzzy details of face image.And aiming at the shortcomings of traditional CS-LBP algorithm in the local information of face image,put forward a new face feature extraction method by integrating the gradient magnitude and gradient phase of the face image,and make use of their complementation to improve face features.At last,the experimental results show that the proposed method is more efficient than the traditional method,and the recognition rate got some increase under different training samples.So the validity of algorithm is proved.(3)Face recognition based on CS-LBP and deep learning.In consideration of the neglect of the local features extraction in deep learing and the large dimensions of facial features in Local Binary Pattern,this paper proposes a facial recognition method based on CS-LBP and deep belief network algorithm.Firstly,the faatures of face images are extracted by CS-LBP operator.Then making CS-LBP texture features as the input of deep belief network,and the good parameters of the network are obtained by layered network training.Finally,it took the experiment on face database and the proved method had better performance than that of face recognition using original methods and methods in other papers.It show that the proposed method has a good face recognition performance.