Detecting data leakage from hard copy documents

Abstract

Document fraud has evolved to become a significant threat to individuals and organizations. Data leakage from hard copy documents is a common type of fraud. This chapter proposes a methodology for analyzing printed and photocopied versions of confidential documents to identify the source of a leak. The methodology incorporates a novel font pixel manipulation algorithm that embeds data in the pixels of certain characters of confidential documents in a manner that is imperceptible to the human eye. The embedded data is extracted from a leaked printed or photocopied document to identify the specific document that served as the source. The embedded data is robust in that it can withstand errors introduced by printing, scanning and photocopying documents. Experimental results demonstrate the efficiency, robustness and security of the methodology.