Merchandise packaging and valuable documents, such as tickets and IDs, are common targets for counterfeit. The deployment of low-cost physically unclonable functions (PUFs) as a way to deter counterfeiting has been receiving increasing attention in both research community and industry.

In this research, we study two types of surface structures: the intrinsic PUFs and the extrinsic PUFs. The intrinsic PUFs explore the optical effect of the microscopic roughness of the surface, such as the paper surface formed by inter-twisted wood fibers. We will briefly discuss the intrinsic paper PUF and show a video demo in this page. The extrinsic PUFs are fabricated by adding ingredient such as fiber, small plastic dots, air, powders/glitters, etc. that are foreign to the surface. We skip the discussion for the extrinsic PUFs and refer interested readers to our academic papers on the topic.

Here, we briefly introduce using an intrinsic, physical feature of paper surfaces for object identification purpose. We used commodity mobile cameras to extract a physical feature of paper surfaces, namely, the surface normal vector field. At the microscopic level, the direction of the surface normal varies from one point to another due to the uneven surface structures created by the intertwisted fibers. This variability in the surface direction can be exploited for the unique identification of a particular patch of a surface.

Fig. 1(a) shows a finger nail sized paper patch and Fig. 1(b) shows the microscopic surface directions for an area of 1/100 of the patch. Such randomness makes one surface unique from another, and can therefore be used to answers a key question of the IoT—the unique addressability issue. An everyday object with a designated “fingerprint” region can therefore be readily integrated into the IoT infrastructure using a commodity mobile phone as a smart sensor. This proposed mobile imaging method tackles the “unique addressability” issue of the IoT at no cost compared to the RFID technology.

Below is a video demonstration for how the proposed technology can be used for countefeit detection by the end-users with mobile cameras. For licensing inquiries, please contact Prof. Min Wu. Dataset is available upon request.