Today, Tim Weyrich, Adam Finkelstein, Nadia Heninger, J. Alex Halderman, Ed Felten and I released our study on Fingerprinting Blank Paper Using Commodity Scanners. This result shows that individual sheets of paper are identifiable by their unique surface texture. The uniqueness of a document’s surface texture is the result of a random deposit of paper fibers during the manufacturing process. By exploiting a property of desktop scanners we are able to accurately and consistently measure the surface texture of a document, extract a fingerprint.

Once extracted, this fingerprint can be digitally signed and printed on the document, or stored in a database. The secrecy of a document’s fingerprint is achieved through the use of a secure sketch. You can think of a secure sketch as a fuzzy hash function. Similar inputs produce identical outputs, but knowing the output tells you provably little about the input. This is ideal for a situation where a document will be subjected to wear and tear. Wear and tear, as well as mis-alignment during scanning, can cause slight variations in a document’s fingerprint, which need to be corrected in order to successfully verify a document.

For more information, please see our website citp.princeton.edu/paper