人脸识别中的核方法

The Kernel Method Applied to Face Recognition

对人脸识别的关键技术和核方法的主要特点进行了简单的分析.指出:将核方法应用于人脸识别中,在特征提取的效率和识别的泛化能力等方面具备技术上的优势.结合国内外相关研究成果,介绍了可应用于人脸识别的三种核方法:基于核的主成分分析(KPCA)、支持向量机(SVM)和基于支持向量的数据描述(SVDD).文章的最后给出了一个应用实例以说明核方法在人脸识别领域中的有效性.

Some key technologies in face recognition and the main characteristics of the kernel method are analyzed briefly. It is technologically advantageous to adopt the kernel method in the process of face recognition in terms of the efficiency of feature extraction and the generalization ability of classification. Combined with the latest research results, three methods based on kernel that can be used in face recognition are introduced: KPCA, SVM and SVDD. At the end of the paper, an example is given to show the validity of the kernel method in the application of face recognition.