Abstract: A novel approach to face recognition based on the common vector combined with 2-dimensional principal component analysis (2DPCA) is proposed in this paper. The common vector of one class is obtained by face images of the class processed by the Gram-Schmidt orthogonalization to represent the common invariant properties of the class. Recognition results are obtained by 2DPCA procedure and distance test of the difference vectors between the original image and the common vector of the class. Experiments are performed on ORL and Yale face databases and the results indicate that the proposed approach achieves good recognition results.