Abstract:
In recent years, identifying source printers has gained immense popularity for forgery detection. Paper employed for agreements and contracts have greater significance and to detect forgery among these papers validates the source more appropriately. With the advent of time, the document forgery is increasing and identifying source printer of document will result in validating a proper source as well as finding that document is forged or tampered. The proposed study aims to identify source printer of printed scanned document images. We employed a dataset for this study that contains 20 printers from which 13 are laser printers and 7 are ink-jet printers having 1200 document images. The documents also contains graphics, tables and text with different font style and sizes. We investigate features through hand-engineered techniques as well as machine-learning techniques. we also investigate the approaches related to text-dependent and text-independent mode. Development of such forgery detection systems are likely to facilitate the forensics community in analysis of printed scanned documents
