Abstract: KPCA extracting principal component with nonlinear method is an improved conventional PCA. The Kernel Principal Component Analysis (KPCA), is used in face recognition, which can make full use of the high correlation between different face images for feature extraction by selecting the proper kernel function. So KPCA can extract the feature set more suitable in categorization than classical conventional PCA. Based on ORL face database, recognizes correlation coefficients of principal component extracted by KPCA. Experimental results demonstrate that KPCA is not only good at dimensional reduction, but available to get better performance than conventional PCA, of which the correct recognition rate is up to 92.5%.