Facial Recognition Based on Kernel PCA

Feature extraction is among the most important problems in face recognition systems. In this paper, Kernel Principal Component Analysis (KPCA) has been used in feature extraction and face recognition. By the use of integral kernel function, one can efficiently compute principal components in high dimensional feature spaces, related to input space by some nonlinear map. Polynomial kernel was used. The experimental results demonstrate that the KPCA is not only good at dimensional reduction, but also available to get better performance than conventional PCA. The highest rate is 90%.