Copy detection pattern-based document protection for variable media

Abstract

A copy detection pattern (CDP) is a shape (typically a rectangle) filled with pixels of random grey levels and incorporated in a digital document to be printed as a unique physical document. It is a powerful tool to detect copies. Whenever a document is printed and scanned, the grey levels of the pixels intermix in a measurable way. When a document of unknown legitimacy containing a CDP is scanned, the CDP distortion is measured and matched against a pre-determined threshold. If this measure is higher than the threshold, an intermediary scan and re-print most likely occurred. For acceptable accuracy, the print quality needs to be reasonably good; therefore in CDP applications, the printer, the CDP media and the scanner must usually be of known type. However, some applications could greatly benefit from CDP features, but have little control over the media. In such applications, the variance of the measurements, owing to media quality variability, is simply too large for a reasonably conclusive detection, at least with current verification techniques. Thus, the authors propose a print-scan model and a smart attack based on the model that causes the CDP copy detection to be mostly ineffective. To handle such an attack scenario and allow a range of different media, new CDP detection metrics were proposed. Their efficacies were tested for different paper types and different attack scenarios. Experimental analysis shows that the proposed features can be used to significantly improve copy detection accuracy.