Research On Face Recognition Algorithm Based On LBP And CNN

At present,face recognition technology at domestic and foreign is one of the hot research topics in the field of pattern recognition and machine learning,which has attracted many researcher's attention and research,it has been widely used in security protection,intelligent video surveillance,identity authentication and other fields.The face recognition system is mainly composed of face image detection,image preprocessing,feature extraction and classification recognition.How to extract and classify face feature information accurately and effectively is the key problem of face recognition.Based on the local binary pattern and convolutional neural network,this paper research the face feature extraction and recognition.The main research contents as follows:? Discussed the research background and significance of face recognition,introduces the research history and current situation of face recognition and the corresponding face feature extraction algorithms,then analysis its advantages and disadvantages simply.? Research a face recognition algorithm based on LBP and 2DLDA.The face image is divided into 4x4,and the face feature is extracted by the unified pattern operator.The extracted face features are reduced by the 2DLDA algorithm,and used the nearest neighbor classifier for recognition.On the ORL and Yale face database,the algorithm advantages can be seen by comparing PCA,LBP,2DLDA,LBP+PCA and other algorithms.?Research a face recognition algorithm based on LBP and CNN.Deep learning has become a hot topic in today's society,the method of deep learning also provides a theory and foundation for the research of face recognition technology.Using the face image as the input of the convolutional neural network directly,there are have problems such as the large dimension of the face image and the neglect of the local structural features of the face.Firstly,the face image is divided into blocks.Extracting local feature information of face images of each sub-block by using local binary pattern,and then combine the local feature information of each sub-block.Using the combined LBP features as input to convolutional neural networks.After multilayer convolution and pooling.Finally,the classification and recognition are completed by the Softmax classifier at the output layer.Simulation experiments on ORL and Yale database show that the recognition effect of this method is better.