A copy detection pattern , secure graphic or graphical code is a small random or pseudo-randomdigital image which is printed on documents, labels or products for counterfeit detection. Authentication is made by scanning the printed CDP using an image scanner or mobile phone camera. The detection of counterfeits using a CDP relies on an "information loss principle", which states that every time a digital image is printed or scanned, some information is lost about the original digital image. A CDP is a maximum entropy image that attempts to take advantage of this information loss. As producing a counterfeit CDP requires an additional scanning and printing processes, it will have less information than an original CDP. By measuring the information in the scanned CDP, the detector can determine whether the CDP is an original print or a copy. CDPs aim to address limitations of optical security features such as security holograms. They are motivated by the need for security features that can be originated, managed and transferred digitally, and that are machine readable. Contrarily to many traditional security printing techniques, CDPs do not rely on Security by Obscurity, as the algorithm for generating CDPs can be public as long as the key used to generate it or the digital CDP is not revealed. CDPs have also been described as a type of optical physical unclonable function.
Security
The theoretical and practical assessment of the security level of CDPs, in other words the detector's ability to detect counterfeit attempts, is an ongoing area of research:
In, practical recommendations on printing stability, taking into account scanning quality of the detector, and managing the security of printing facilities.
In, a decision theoretic-model to determine optimality properties of CDPs and under an additive Gaussian noise assumption for the print channel and an optimal showed that the optimal decision function is a correlator.
In, different new CDP detection metrics are proposed and confirmed a significant improvement of copy detection accuracy.
In, the impact of multiple printed observations of the same CDP is studied, and it is shown that the noise due to the printing process can be reduced but not completely removed, due to deterministic printing artefacts.
In, a theoretical comparison is made between the performance of CDPs and natural randomness.
In and, deep learning methods are used to recover portions of the digital CDP, and it is shown that these can be used to launch clonability attacks.
In, quality control challenges are reviewed, and an inline verification system of secure graphics is proposed for high security printing applications.
In, different attack methods based on restoration of the scanned CDP are tested. and show that a classifier based on support vector domain description outperforms other classification methods.
Applications
CDPs are used for different physical item authentication applications:
and digital watermarks are inserted into banknotes to be detected by scanners, photocopiers and image processing software. However the objective of these techniques is not to detect whether a given banknote is a counterfeit, but to deter amateur counterfeiters from reproducing banknotes. Digital watermarks can may be used as well to differentiate original prints from counterfeits. A digital watermark may also be inserted into a 2D barcode. The fundamental difference between digital watermarks and CDPs is that a digital watermark must be embedded into an existing image while respecting a fidelity constraint, while the CDP does not have such constraint.
