Research On Facial Expression Recognition Methods Based On Multi-features Fusion

With the continuous development and perfection of artificial intelligence, facial expression recognition is becoming an important research direction in the field of human-computer interaction and artificial intelligence, has profound theoretical significance and application prospects. Implement the computer’s facial expression recognition will promote the development and application of artificial intelligence technology, and even play a important role in the development of psychology and other disciplines in future.By studying and learning a large number of domestic and international research papers and literature, we discussed and researched a number of issues in facial expression recognition, and for a variety of expression recognition based mutli-features fusion for a more in-depth study. This paper describes the background and applications in the field of facial expression recognition firstly; And then summarizes the current status of research expression recognition; Thirdly,propose a method of decision-level fusion methods and feature level fusion, and verified the proposed two methods of experimental. The research work in this paper mainly includes the following several respects:（1）.We propose a facial expression recognition method with discriminative multitask joint sparse representation based on local image texture feature and global coordinate information in this paper. To solve the problem that the combination of multi-features are linear. We propose a facial expression recognition method with discriminative multitask joint sparse representation based on local image texture feature and global coordinate information in this paper. First, Texture features of the image and global coordinate information are achieved to construct training dictionary, discriminant loss function are introduced to optimize the dictionary. Second, introduce a joint sparse Regularization term in objective function to joint sparse representation. At last,the sparse coefficient matrix after sparse coding can be used for training and expression classification in SVM. The proposed method is extensively evaluated on Curtin Faces Database and BU₃DFE Database. The experimental results show that this method can reduce feature dimension and mining the correlation between multi-features, and enhances the power of discrimination of Sparse coefficient matrix. In compared with the state-of-the-art methods in expression recognition, the proposed method has about 2.5%-5% improvement on the BU₃DFE Database.（2）.We proposed a face expression recognition based local discrimination and cost-sensitive Adaboost. Aiming at the problem of the uncertainty of the results of single classifier we proposed a face recognition method based on local discriminant and cost sensitive Adaboost. This method can be divided into two stages, the first stage is the selection stage of the weak classifier. Calculating each training sample’s local discrimination factor in each round of iteration, select the largest local discrimination factor as discrimination factor in this round, and select the corresponding weak classifiers are used as the current optimal weak classifiers. The second stage is the weight updating stage. Introduce the cost-sensitive loss function to minimize the misclassification cost to obtain the weak classifier in each round of iteration, and update the training sample distribution weights. Finally, use the strong classify recognize facial expression of these samples. Experimental results on BU₃DFE show that this method can effectively make the traditional Adaboost to solve minimize the error rate to minimum error classification cost, and improve the performance of Adaboost algorithm to improve the accuracy of classification.（3）.A prototype system of facial expression recognition methods based on multi-features fusion is designed and implemented. This system adopts the object-oriented design idea and the mixed programming model of MATLAB and VC#. The system implements eight modules including face area recognition, landmark marked, face image pre-treatment, feature extraction, dictionary learning, feature fusion, sparse representation, Adaboost algorithm, and designs a user-friendly interface. This system is used to illustrate the availability of the algorithm.