Face Recognition Algorithm Using The Wavelet Transform And The Sparse Coding

With the development of modern science and technology, many sectors are increasingly high requirements for security, face recognition has been widely applied to many fields, because that is not need to contact to collect information, as well as the advantages of high safety factor, such as, the face recognition attendance system, the face recognition access control systems. The face recognition system is mainly consists of face detection and recognition process. The main research of this article is the recognition stage, the stage includes feature extraction and classification in the recognition process. Whether the feature extraction and classification algorithms is accurate or not, it is will have an reflect on the speed or the recognition rate of the face recognition. So it is important to choose the feature extraction algorithms and the classification algorithms.In this paper, we choose the non-negative sparse coding algorithm in the feature extraction stage, the algorithm uses the idea of sparse, that can extract the local features of face images. Compared with the principal component analysis (PCA) algorithm, this algorithm can improve processing speed and reduce the storage space. In order to accelerate the speed of face recognition, in this paper, we are use of the frequency localized features and multi-resolution characteristics of wavelet transform before feature extraction. The algorithm can remove noise and retain the local features of face image. Finally, we do a large number of experiments, it is found that non-negative sparse coding has good adaptability to the block face. In the classification stage, this article select the approximation of support vector machine algorithm, the algorithm can be used to multi-class classification, so the algorithm has practical value.