A Face Recognition Method Based On Improved LDP Features

Face recognition has the advantages of easy operation and stable performance in the field of authentication,and it has been widely used in recent years.Higher requirement has been put forward on the accuracy and the running speed of face recognition in this situation.Based on the basic principles of face recognition and the study of domestic and foreign literature,this paper completes the study of image preprocessing,face detection,feature extraction,complete recognition realization and result verification,and improves the method of face recognition.Firstly,this paper summarizes the development of face recognition algorithm and determines the research direction based on the local feature method.Secondly,histogram equalization,spatial domain filtering and image sharpening are introduced respectively,and the research of image preprocessing is finished.Then the Adaboost algorithm is studied,and an effective face detection method is validated by using the Haar feature classifier,which is provided by the OpenCV platform.The integral graph and the cascade classifier model are introduced to optimize the face detection speed.After that,the local feature extraction methods including basic LBP,improved LBP and LDP are studied,and an improved LDP feature extraction method is proposed.The method includes the following steps:cutting the face image into sub-regions of same size;taking the negative gradient returning value into consideration when the feature code of the center pixel is obtained;Different weight values are assigned to each sub-region according to the difference of their structural contrast information.Finally,the face recognition performance of LBP,LDP and improved LDP is verified and compared on ORL and Yale face database.The weighted matching of sub-region histogram ensures different sub-regions play different roles in the processing of feature histogram matching.Compared with the traditional LBP and LDP methods,the face recognition method based on improved LDP features has higher success rate of face recognition,which means that this improved scheme is effective.