The optical duplication of physical keys via decoding of bitting patterns from images is a known threat. However, since the features being analyzed are small and differences between each bitting value amount to only a few pixels, the stealthy application of this attack in practice is limited by the difficulty of obtaining clean high-resolution imagery of a target key. We use image processing techniques to overcome the common drawbacks of surreptitious or field-captured imagery. Superresolution sensing can be used to reconstruct high-resolution images from a series of low-resolution images from mostly-redundant viewpoints. Furthermore, related image processing models can, with or without the availability of motion data from the sensor, overcome motion blur (e.g. from smartphone video captures). Results presented in this talk will demonstrate that a stealthy optical key duplication attack can be plausibly performed without sophisticated equipment or logistical support. We will also consider the implications for entirely remote, opportunistic attacks.
Ian is a PhD student at North Carolina State University