The growing availability of counterfeit drugs has greatly increased the need for track and trace of drugs through the supply chain. Although numerous methods such as tagging, labelling, and marking are available for identifying and authenticating drug packages, none are available for the contents. A method is presented for directly identifying individual medicine tablets without any markings or additives. A lighting system is used to capture an image of a tablet’s micro-scale surface bumps, which is then used as the "fingerprint" of that tablet. Since plain unmarked tablets have no macroscopic (humanvisible) feature, the orientation of the fingerprint image cannot be controlled during capture. Tablets are thus identified by geometric invariant image matching using Fourier-Mellin phase features. Experimental results show that the proposed image matching algorithm enables accurate and robust identification of numerous plain tablets.