FIBAR: Fingerprint Imaging by Binary Angular Reflection for Individual Identification of Metal Parts

This paper proposes an imaging method for individual identification of metal parts using the image features of the appearances as ``fingerprints´´. The key to put such a method into the real applications is how to capture the unique features that are ``repeatable´´ across varying environments. Since the image features of metal surface are quite unstable due to specular reflection, most previous methods require special imaging devices. In this paper, we propose an imaging method, named Fingerprint Imaging using Binary Angular Reflection (FIBAR), for capturing repeatable ``fingerprints´´ from metal surface by using a common camera. FIBAR captures an image of the angular black/white intensity pattern that is reflected off the metal surface. Since FIBAR provides a common camera with plenty of repeatable ``fingerprints´´, the registration and query images of the same part are robustly matched under varying environments. FIBAR is implemented as an inexpensive tool by 3D printing, and it is distributable for wide applications. In our experiments, 1,000 metal bolts with identical appearance manufactured with the same mold are perfectly identified by employing FIBAR.