A Occluded Face Recognition Method Based On Sparse Representation

Face recognition is a kind of based on people’s facial features information identification of biological recognition technology. With a camera or camera collection contains face images or video stream, and automatically detecting and tracking human face in the image and then to detect the face to face a series of related technologies. There have a lot of Biometric technology such as speech recognition, fingerprint recognition, iris recognition and palm print recognition. Because the feature of face recognition is natural, friendly and less disturbance to the user and the advantages of easily accepted by users, it has a broad application prospect in the information society for indispensable identification method. In recent decades, the research of face recognition which based on two-dimensional images has made great progress. The system of face recognition has good recognition performance under constraint environment. But in practice, face recognition system for face image acquisition is under the environment which without constraint, in the shadow of the people or objects, these interference factors make the face of incomplete data, thus affecting the entire face recognition system accuracy. The key problems of us is that how to handle the face with blocks, this paper mainly studies the image-based shade face mask removal and the identification method, the main research results include:Firstly, we summarize the development history and research significance of face recognition, facial recognition technology development. The existing 2D face recognition method based on the characteristics of the extracted features and matching method of different classification summary. Mainly introduces the existing condition of 2D face recognition methods and their characteristics.Secondly, we introduce the idea of sparse expression, the methods of dictionary learning and the methods of the basic classification. Several kinds of occluded face recognition based on sparse express processing method were introduced.Finally, the algorithm based on the sparse classification method is put forward to remove occlusions from frontal facial images. Occluding objects in facial images may greatly degrade the recognition performance and robustness of face recognition system. The essence of sparse express is to use the least amount of training samples to represent the testing sample images. In this paper, the method of remove barrier is compute the sparse coefficient on the training set without sunscreen, then reconstruction the image without occlusion.