Recognition of planar objects under full variations in viewpoint: Invariants versus mental transformations