A common strategy for extracting the feature and to preserve the global structure such as Principal Component Analysis, Two Dimensional Principal Component Analysis and Linear Discriminant Analysis

A common strategy for extracting the feature and to preserve the global structure such as Principal Component Analysis, Two Dimensional Principal Component Analysis and Linear Discriminant Analysis have been used. These schemes are a classical linear technique that projects the data along the directions of maximum variance. To improve the performance, Locality Preserving Projection is used. The objective is to preserve the intrinsic geometry of the data, and local structure. However, Locality Preserving Projection has the weakness, restrictiveness to separate the non linear data set. A novel approach to separate non linier data set based on selection of non linier function maximum value by using Kernel is proposed. Kernel maps the input to feature space by using three non linear functions; the result of mapping will be selected the maximum value. To avoid singularity, the result of the selected value will be processed by using Principal Component Analysis. Furthermore, Laplacian is used to process the result of Principal Component Analysis to achieve the local structure. The performance of the proposed method is tested to classify smile stages pattern. The experiment result shows that, the proposed method has higher classification rate than Two Dimensional Principal Component Analysis and combining of Principal Component Analysis, Linear Discriminant Analysis and Support Vector Machine.