Font Size: a A A

Based On Multi-class Support Vector Machines Face Recognition

Posted on:2011-12-03Degree:MasterType:Thesis
Country:ChinaCandidate:Q M LuFull Text:PDF
GTID:2178360305982248Subject:Computer Science and Technology
Abstract/Summary:PDF Full Text Request
Face recognition is a new biometric technology, fingerprint recognition, iris recognition is similar to the identity as the only important means of authentication. Face to portrait the basis for analysis, just use cheap capture devices. Portrait access is easy to operate, easy to fraudulent use of the face, the face can carry and will not be lost, to prevent arrived lazy and so on. Therefore, the field of face recognition security concerns as one of the hotspots. Support vector machine is based on statistical learning theory and the theory of VC dimension structural risk minimization an important theoretical basis. As a new machine learning method, SVM can solve the small sample, nonlinear, high dimension and local minima, the actual problem. SVM classification algorithm will be applied to speaker recognition, pattern recognition, can be effectively solved through traditional classifier learning, generalization error and the curse of dimensionality problems.Efficient extraction of image texture features are used on the following support vector machine classifier learning and training have a very important role. The problem of image feature extraction, the contents of this thesis, the image texture features in a variety of description and extraction. Texture features from the actual application, the focus of the statistical method in the gray co-occurrence matrix algorithm.Since SVM has a solid theoretical foundation and good classification performance advantages, such as support vector machines in the statistical theory and classification theory based on the principle, from the feature vector extraction, nuclear function, training algorithm and multi-class classification algorithm for three important of recognition performance and speed of the aspects of research and analysis, this thesis presents a support vector machine and the image texture characteristics of the face recognition method and framework of the model. First, the image preprocessing, such as the gray level transformation, histogram equalization, image smoothing, and then use texture features technology from the processed image can be used to extract support vector machine training feature vector, the last use of Support Vector Machine class classifier, the training and test samples, the classification for face recognition.In this thesis, to establish their own sample libraries on the basis of my face, the face image to a support vector machine classification experiments. Experimental results show that the method can effectively improve the recognition rate of face recognition systems. This finally made some meaningful research results, for the face recognition study provides strong data and positive suggestions.
Keywords/Search Tags:Face Recognition, Support Vector Machine, Texture, Statistical Learning Theory, Preprocess
PDF Full Text Request
Related items