Support Vector Machine In3D Face Recognition

Support vector machine (SVM) is a new machine learning method developed in the framework of statistical learning theory, which bases on the VC-dimension (advanced by Vapnik and Chervonenkis) theory and the principle of structural risk minimization rule. It aims at finding optimal classified hyperplane. The optimal classified hyperplane should classify the two types of objects rightly, and in the meantime makes sure the margin between different classifications is the largest.In this paper, profound research is made on the algorithm of One-class SVM, and it is used in the classification of3D face recognition. The main contents of this paper are as follows:First, the reasons is analyzed that when in the algorithm of One-class SVM different sizes of training sets are unbalanced; the results will be inclined to the larger one. This algorithm is improved by raising weights of small sample data, which lowers the impacts caused by the different sizes of training sets and reduces errors of the training model.Second, weighted One-class SVM classifiers is applied to3D face recognition. In the beginning, the eigenvectors of face images are extracted by PCA, then the eigenvectors are made as the input vectors of One-class SVM, in addition, the face images are trained by the weighted One-class SVM, which generate weighted One-class SVM classifiers. Furthermore, the algorithm of weighted One-class SVM is used to recognize the face images. In the end, the experiments on3D face database show the feasibility of the algorithm.