Efficient representation of in-plane rotation within a PCA framework

In this paper, we derive an analytic representation of the eigenvectors and coefficients for in-plane rotated images. This, on the one hand, allows an efficient PCA-basis calculation in the learning stage and on the other hand a direct computation of the rotation angle in the recognition stage. In the experimental section, we demonstrate that the new method is feasible and the recognition and out-of-plane results are comparable to the standard method.