Font Size: a A A

Face Recognition Study Based On Kernel Principal Component Analysis And Support Vector Machine

Posted on:2011-01-06Degree:MasterType:Thesis
Country:ChinaCandidate:B C WangFull Text:PDF
GTID:2178330332460687Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
With the development of technology, biometric technology has become an important way of the personal identification or authentication technology, which has been a very active research topic. Recognition, as an important branch of biometrics, which is non-invasive and most natural for users, has been the most intuitive means of identification for its more easily accepted in information security, criminal detection, access control and other areas, and has showed a wide range of applications.A new approach for face recognition based on kernel principal component analysis (KPCA) and support vector machine (SVM) is presented to improve the recognition performance of principal component analysis (PCA), which has shortage in processing nonlinear image problems. Meanwhile, this method can be applied to solve both overfitting problem and small sample problem. The KPCA method is performed on every facial image of training set to get kernel facial features of training samples. Under the principle that the information of image loses as little as possible, the face data of high dimension feature space is shadowed into low dimensional space. The SVM face recognition model is established and then the low-dimensional space information data is identified. Experiment results demonstrate that the approach proposed in this paper is efficient.To verify the validity of the method, the ORL face database is used for face recognition experiments. Through the different contribution of KPCA and the number of training samples, and in comparison with the previous face recognition methods, the results show that the proposed method has better recognition effect.
Keywords/Search Tags:Machine Learning, Face Recognition, KPCA, SVM, Kernel Function Methods
PDF Full Text Request
Related items