In subspace pattern recognition, the basis vectors repff!sent the features of the data and define the class. In the previous }1-Vrks, standard principal component analysis is used to derive the basis vectors. Compared with standard PCA, Nonlinear PCA can provide the high-order statistics and result in non-orthogonal basis vectors. We combine Nonlinear PCA and a subspace classifier to extract the edge and line features in an image. The simulation results indicate that the basis vectors from Nonlinear PCA can classify the edge patterns better than those from linearPCA