Model based print signature profile extraction for forensic analysis of individual text glyphs

Forensic analysis of individual printed items, including single characters, enables the addition of some level of security to any printed item (label, document, package, etc.). In this paper, we present a model-based approach for extracting a signature profile around the outer edge of virtually any text glyph. We show that for two high-resolution imaging devices (the Dyson Relay CMOS Imaging Device, called DrCID, and a high speed line-scan camera) this signature encodes that part of the glyph boundary that is due to the random fluctuation of the print process, enabling significantly higher levels of forensic discrimination than previously shown. The model-based approach enables a security workflow where the line-scan device is integrated into production line inspection with later forensic investigation in the field using the DrCID device. We also develop a simple shape descriptor to encode the signature profile, making it easier to manipulate, test and store. We argue that the shape descriptor provides forensic-level authentication of a single printed character.