Eigenfaces are eigenvectors. When performing principal component analysis (PCA) on a set of images showing human faces, the principal directions or the components that span the space are called the eigenfaces instead of eigenvectors because the underlying data are faces... its just another name.