Research And Implementation Of Face Recognition And Expression Recognition Based On Human Face Biological Characteristic

With the development of computer technology and artificial intelligence technology,biometrics technology has become more and more widely used in national defense security,social security,commercial finance and other fields.It has gradually become a key technology in government,society and commerce.In biometrics,facial analysis based on human face features such as face recognition and facial expression recognition is an important research and development direction.In particular,it has developed rapidly in recent years and has extensively studied in the fields of pattern recognition,computer vision,artificial intelligence and human-computer interaction.So far,the face analysis technology based on face biological features has integrated the knowledge of many subjects such as computer vision,artificial intelligence,human-computer interaction,deep learning,software design,electronic hardware and so on,and has been widely involved in many current frontier fields technical knowledge.In this paper,face recognition and facial expression recognition are studied based on biological features of human face.Based on practical application,face recognition and facial expression recognition based on camera video are realized.In this paper,face recognition of eigenface based on PCA dimensionality reduction and expression recognition of SVM method based on ASM model are proposed.The main research contents of this dissertation are as follows:(1)Face detection.Face detection is a very important preparatory work in the process of face recognition and facial expression recognition,and it is necessary to carry out the work.In this paper,a feature-based method is used to implement face detection.Based on Harr-like features,LBP features and MB-LBP features of face images,face detection is implemented by using Adaboost classifier.The face detection based on the MB-LBP feature can detect and detect the fast face detection with the change of expression,complex lighting changes,complex background and face angle change accurately.(2)Image preprocessing.Image preprocessing is an important work in recognition classification.In this dissertation,the methods of grayscale normalization,size normalization,median filtering denoising,and gray histogram equalization enhancement are used to preprocess human face images.These image preprocessing methods can well eliminate the influence of various factors such as illumination,noise,brightness change and size change on the image recognition.Among them,the median filtering denoising achieves good noise reduction,Image equalization to achieve the image enhancement processing to make the details of the image features more clearly.(3)Face recognition classification.In this paper,a face recognition method of eigenface based on PCA dimensionality reduction is proposed to classify face recognition.All face features are reduced to the average face by PCA,and the face feature space is constructed by covariance matrix.Finally,Recognition of face features and facial features of the distance between the space for classification and identification.(4)Landmark detection and location.The detection and positioning of landmark mainly locate and locate the facial feature points such as eyes,nose,mouth,eyebrows and facial outline in the face area.In this paper,two landmark detection and localization methods based on ASM model and AAM model based on MB-LBP feature are researched,and the detection and localization of 68 key feature points on face are successfully achieved.(5)Facial expression recognition classification.In this paper,the SVM method based on ASM model is implemented to recognize facial expression.The ASM model method is used to extract the eigenvalues of 68 key feature points of facial expression facial expression.SVM classifier algorithm is then used to train the eigenvalues.Decision tree thinking expression recognition classification.(6)Face recognition and expression recognition experiments.Using the standard face database and expression database for experiments,this paper validates the research methods,statistics of the face recognition rate and expression recognition rate of the method,and compared with the recognition rate of the method in the literature,and found that the method of Advantages and disadvantages,the method of optimization of this article.